| Age | Commit message (Collapse) | Author |
|---|
|
This uses the new libdrm busy interface, once I had this in place
I added a error if this happened and it does on my desktop here,
so may as well add the optimisation that used to be in my old KMS tree.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
Still haven't implemented the actual watermark
calculation, but this forces the display requests
to urgent if the user specifies DisplayPriority HIGH.
|
|
This requires an X server from git with createpixmap2 support fixed up in it.
On 1.6 and previous it won't do any tiling, if you are running git server
please upgrade to latest git.
Option "AllowColorTiling" "true" to enable and do some testing
|
|
If the DisplayPriority option is set to HIGH,
force the diplay mem requests to urgent.
|
|
We should print this after DAC detection to be useful for TV
|
|
For some reason we lost the DACB source when we dpms tv back on.
|
|
We end up writing back an empty save register over the dac
detection results so the atombios table can no longer
figure out the S-video vs composite so sets the wrong one up.
this gets tv-out on my r580 with a composite connector working.
|
|
If we need to use the second scaler actually set the second scaler
up.
|
|
|
|
for rect textures you don't want to use wrap clamping which
is the default.
|
|
|
|
from Hartmut Niemann
|
|
Abstract vertex buffer, preliminary work before others change
to bring cs support along IB support.
|
|
|
|
Also remove a superfluous assignment.
|
|
|
|
Check we can handle the bpp, and remove superfluous flush.
|
|
Fixups by Michel Dänzer: Doesn't seem to be a win in terms of raw numbers, but
should allow more pipelining and avoiding BO memory waste for small glyph
pixmaps.
|
|
a) turn of R200_RE_CNTL - SCISSOR_ENABLE - this save us emitting R200_RE_TOP_LEFT, note scissor is still enabled.
b) disable aux scissors.
|
|
|
|
fixes bug 19943
|
|
What we want to do is add userspace object support to radeon
kernel modesetting. Also this DFS is dumb and might endup doing
blit from GTT to GTT.
|
|
|
|
|
|
Use XAA in low memory situations or when the DRI is disabled.
Using shadowfb might also be a viable option, maybe even a better option...
fixes bug 21683
|
|
fixes bug 23065
|
|
fixes bug 23064
|
|
fixed bug 23063
|
|
fixed bicubic filtering after VT switch or suspend/resume
|
|
From RHEL QE testing, we could end up with the cursor at 0 since
we think EXA is in use when really it isn't. the info->useEXA = FALSe
might be unnecessary but better to be explicit
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
Should fix bug 19943
|
|
|
|
|
|
Without this, apparently they are initially allocated in system RAM (or
possibly GTT), so any GPU rendering to them first triggers a copy of the whole
uninitialized data to VRAM...
We may want to explore more sophisticated schemes in the future, but for now
this seems to improve KMS 2D performance quite a bit.
|
|
- r1xx-rv350 chips have the old pci gart
- rv380+ chips have newer pcie gart
Select the right kind regardless of whether the user selects
PCI or PCIE.
|
|
We need to map front buffer and also to avoid EXA initialization
when falling back to shadowfb on newer r6xx/r7xx when KMS is enabled.
|
|
Also, r6xx/r7xx don't have indirect MC space, so
don't try to access it.
|
|
|
|
Fixes build against older xserver.
|
|
The LeaveVT hook isn't always called when the server dies, e.g. when quitting
from the GDM greeter. This may cause existing servers to fall over if the dying
server process still exists when they try to re-acquire master and set a mode.
Also use drmSet/DropMaster() rather than ioctl() directly.
|
|
Otherwise some things like changing the colour map won't work properly.
|
|
As recommended by the register reference when using the WAIT_CRTC_VLINE bit in
the WAIT_UNTIL register, as we are.
|
|
|
|
Most of the default state setup was extraneous or duplicated
in the accel ops. There's no need to reset all the consts
everytime you reset the default state for example. This leads
to a 10x reduction in default state.
|
|
DPMS header was split into dpms.h (client) and dpmsconst.h (server). Drivers
need to include dpmsconst.h if xextproto 7.1 is available.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
|
|
This causes relocs to be emitted for the colorpitch.
Shouldn't have no effect on current kernels, but will keep DDX
going on kms kernels when tiling lands
Also contains a missing reloc space for r200
|
|
fixes bug 22726
|
|
Upload the bicubic texture once during textured video init
rather than once per frame. Suggested by Michel Daenzer on
IRC.
|
|
fixes bug 22730
|
|
Useful to building with mixed kms/non-kms setups
|
