Dave Airlie
7677fd4765
radeon: fixup space accounting + add assert for now
2009-04-06 09:46:47 +10:00
Ben Skeggs
62888f088f
nouveau: idle the channel a bit better before destroying it
2009-03-25 15:00:52 +10:00
Ben Skeggs
8c46fa5a40
nouveau: fix potential oops in gpuobj_channel_takedown
2009-03-25 15:00:20 +10:00
Ben Skeggs
1e7d005aa0
nouveau: prevent fbcon notifier waits when unnecessary
...
It seems fbcon_sync will get called continually by something even while
not at a console, so ignore the call unless we've previously rendered
to the fbcon.
2009-03-25 14:59:07 +10:00
Ben Skeggs
f4cb7a7340
nv50: turn off crtcs in display_disable
2009-03-25 14:57:25 +10:00
Dave Airlie
41e736c795
radeon: add name function for speeding up dri2
2009-03-23 18:32:36 +10:00
Ben Skeggs
47c68ee6e4
nv50: enable dithering by default on lvds
2009-03-19 16:11:06 +10:00
Ben Skeggs
56887e62e7
nv50: some kms fixes
2009-03-19 16:11:05 +10:00
Ben Skeggs
4fec64f166
nv50: unpin cursor when not bound to a crtc
2009-03-19 16:11:05 +10:00
Ben Skeggs
abfe1738e0
nv50: pin/unpin framebuffer as required
2009-03-19 16:10:59 +10:00
Ben Skeggs
7e0ed73e0c
nouveau: use nouveau_fence_channel() to get chan for m2mf move
2009-03-18 19:30:42 +10:00
Ben Skeggs
ec5abde7b3
nv50: kill off nv50_cursor
2009-03-18 19:21:57 +10:00
Jakob Bornecrantz
06d9eff137
mode: Make xfdrmMode.[c|h] not depend on drm_mode.h
...
(cherry picked from commit eb78c53aa1
2009-03-13 18:44:08 +10:00
Ben Skeggs
d6f6108ad0
nv50: add update flag to set_dither()
2009-03-11 22:48:34 +10:00
Ben Skeggs
aeae747a69
nv50: more reordering to avoid display hanging on some chipsets
...
These things are so touchy!
2009-03-11 22:48:08 +10:00
Ben Skeggs
3fc105f00a
nouveau/ppc: no instmem access at BIOS shadow time
2009-03-10 20:17:52 +10:00
Ben Skeggs
cce3eb4e26
drm/nv50: assorted kms fixes/cleanups
2009-03-10 18:22:42 +10:00
Ben Skeggs
16122cb1d9
drm/nv50: silence ddc timeout messages unless debugging enabled
2009-03-10 18:15:55 +10:00
Ben Skeggs
cbbbc9a1fc
nv50: remove division from nv50_vm_bind_linear
2009-03-09 21:22:59 +10:00
Ben Skeggs
15e2184f74
nouveau: respect the nomap flag for gem objects
2009-03-09 11:04:49 +10:00
Ben Skeggs
fa73fa6889
nouveau: big rewrite of nv50 kms code
...
This is a essentially a big squash merge from the kernel tree I've been
working in, sorry about the loss of history, but commiting each patch
individually would've taken quite a while to do.
The nv50 kms code now uses the helpers in the drm, simplifying the
code greatly.
The code is far more stable than previously on all the G8x cards I've
been able to test on, but there's likely still bugs to be ironed out!
2009-03-09 10:58:57 +10:00
Ben Skeggs
f158b0253b
nouveau: disable/enable PGRAPH FIFO access around channel creation
2009-03-09 08:07:55 +10:00
Ben Skeggs
54a66c3b44
drm/nouveau: implement NVOBJ_FLAG_ZERO_FREE
...
Not really necessary it seems as we've been dealing without it for a while
now. But, it's useful to be sure that we don't end up with stale bits
of state laying around.
2009-03-09 08:06:30 +10:00
Stuart Bennett
6f76c1480c
nouveau: don't try to traverse non-existent lists
...
Fixes nouveau_ioctl_mem_free Oops
2009-03-06 09:09:14 +10:00
Ben Skeggs
96e09e22a6
drm/nouveau: make portion of vram as reserved for PRAMIN on all chipsets
...
NV04 was completely busted. Push buffers were getting allocated at the
end of VRAM, overwriting PRAMIN. So, it turns out PRAMIN is in VRAM on
all chips. Question answered!
2009-03-06 09:09:14 +10:00
Ben Skeggs
6e68f1f2b0
drm/nouveau: remove gpuobj ref from list even if not found in ramht
2009-03-06 09:09:14 +10:00
Ben Skeggs
baf814d1f2
nouveau: unbreak <nv50 and any ppc cards after instmem changes
2009-03-06 09:09:08 +10:00
Ben Skeggs
7946b7615f
nouveau: fix oops in gem_new() failure path
...
It turns out drm_bo_usage_deref_locked doesn't set the pointer it's passed
to NULL after destroying the object. So, this double free made things go
boom!
2009-03-06 09:09:08 +10:00
Maarten Maathuis
c921f402b7
nv50/kms: fix crtc timings to match ddx
2009-02-27 14:25:47 +01:00
Thomas Hellstrom
86a5f08fdb
drm/ttm: work around some x86 pat oddities by using mixedmap instead of pfnmap
2009-02-27 14:01:33 +01:00
Dave Airlie
cc2b7ffc85
radeon: do not use this flushing mechanism yet.
...
This flush seems to cause instability at the moment,
I can run glxgears on a cube now a lot better without this
2009-02-23 14:36:40 +10:00
Ben Skeggs
b283b22d0b
nv50: leave vm offset 0-512MiB invalid
...
To catch "NULL pointer" access
2009-02-22 12:54:24 +10:00
Ben Skeggs
e033cbf09a
nv50: use nv_wait for PRAMIN page table update flushes
2009-02-22 11:40:12 +10:00
Ben Skeggs
f38b654609
nv50: remove redundant instmem flush after PRAMIN page table update
2009-02-22 11:40:08 +10:00
Ben Skeggs
9ab2a667b9
nv50: de-typedef nv50_instmem_priv
2009-02-22 11:40:02 +10:00
Ben Skeggs
05cda72eb7
nouveau: fix backwards conditional in wait_for_idle()
...
oops..
2009-02-22 11:39:48 +10:00
Ben Skeggs
569d3f8281
nv50: do instmem flush magic on finish_access() after writes
2009-02-22 11:39:43 +10:00
Ben Skeggs
afac6a4ddc
nouveau: hide instmem map, force use of {prepare,finish}_access
2009-02-22 11:38:41 +10:00
Ben Skeggs
cdba1bf2e4
nouveau: decapitalise dev_priv->Engine
2009-02-22 11:37:17 +10:00
Ben Skeggs
5a28449f88
nv50: remove clamp to 512MiB VRAM
2009-02-20 15:33:29 +10:00
Ben Skeggs
8616e8499b
nv50: rework vm handling so we can support >512MiB VRAM
2009-02-20 15:26:24 +10:00
Ben Skeggs
3c649c9329
nouveau: generalise waiting on register status with timeout
2009-02-20 15:23:50 +10:00
Ben Skeggs
393494d9bd
nv50: fix some fbcon corruption issues
2009-02-20 15:21:42 +10:00
Ben Skeggs
6ec57ecae6
nouveau: signal any outstanding fences on channel destroy
2009-02-20 15:04:23 +10:00
Ben Skeggs
32a7de52a7
nv50: basic fbcon acceleration
2009-02-20 15:03:07 +10:00
Ben Skeggs
776d4fe697
nouveau: use consistant register access macros
...
We had quite a variety, name them all similarly and add a few new
ones that'll be needed for modesetting.
2009-02-20 14:57:40 +10:00
Ben Skeggs
bedffcd3bc
nouveau: always map mmio regs at load time, not first open
2009-02-20 14:39:52 +10:00
Pekka Paalanen
54ba25571a
nouveau: support backlight only when kernel does
...
Loading nouveau.ko would fail with unknown symbols, if the backlight
class device support is not provided in the kernel. Let's make the
backlight support dependant on the kernel configuration.
This is a bit ugly, the proper way would be to check for the config in
Makefile.kernel whether to build nouveau_backlight.o at all, and if not,
nouveau_drv.h should provide the stubs.
Signed-off-by: Pekka Paalanen <pq@iki.fi>
2009-02-20 14:17:02 +10:00
Matthew Garrett
ba41d2df3f
nv40, nv50: fix backlight build for <2.6.29 kernels
...
Signed-off-by: Matthew Garrett <mjg59@srcf.ucam.org>
Signed-off-by: Pekka Paalanen <pq@iki.fi>
2009-02-20 14:16:54 +10:00
Ben Skeggs
61ff524373
libdrm/nouveau: free drmVersion after we're done with it
2009-02-20 09:25:35 +10:00
