Ben Skeggs [Fri, 16 Nov 2012 00:24:31 +0000 (10:24 +1000)]
drm/nvd0/disp: start implementing support for older display classes
Currently unused, but checkpointing the merged head handling routines.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 15 Nov 2012 08:58:01 +0000 (18:58 +1000)]
drm/nouveau/dp: remove last bits of VBIOS parsing from DRM code
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 15 Nov 2012 08:56:02 +0000 (18:56 +1000)]
drm/nouveau/dp: move core link training calls to common code
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Sun, 11 Nov 2012 23:55:16 +0000 (09:55 +1000)]
drm/nouveau: remove (now obsolete) BIT U table parsing from DRM code
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 9 Nov 2012 02:09:48 +0000 (12:09 +1000)]
drm/nv50/disp: move remaining interrupt handling into core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 9 Nov 2012 01:25:37 +0000 (11:25 +1000)]
drm/nvd0/disp: calculate U script id in supervisor interrupt
This is like we do on nv50:nvd9 already. There's been no problems seen
yet with using this *seemingly* scratch register to store the value, but
we won't be able to do this anymore once nv50's code is merged.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 9 Nov 2012 00:54:38 +0000 (10:54 +1000)]
drm/nv84/disp: move hdmi control into core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 9 Nov 2012 00:38:10 +0000 (10:38 +1000)]
drm/nva3/disp: move hda codec handling to core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 23:53:28 +0000 (09:53 +1000)]
drm/nv50/disp: move dp link training helpers into core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 23:38:06 +0000 (09:38 +1000)]
drm/nv50/disp: call into core for dac load detection
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 23:32:56 +0000 (09:32 +1000)]
drm/nv50/disp: call into core to handle dac/sor power state changes
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 23:22:31 +0000 (09:22 +1000)]
drm/nv50/disp: allocate display from driver core
EVO channels still handled "manually", this won't be ported here, and
will instead be held off until nv50_display/nvd0_display are merged.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 07:41:06 +0000 (17:41 +1000)]
drm/nouveau/bios: implement some dcb output entry parsing/matching functions
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 04:59:26 +0000 (14:59 +1000)]
drm/nvd0/disp: handle DP transfer unit setup from second supervisor interrupt
This is what we've done forever in nv50_display.c, and also allows the
last direct MMIO accesses to be removed from nvd0_display.c.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 04:22:28 +0000 (14:22 +1000)]
drm/nvd0/disp: move HDMI control to core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 04:03:56 +0000 (14:03 +1000)]
drm/nvd0/disp: move HDA codec setup to core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 02:08:55 +0000 (12:08 +1000)]
drm/nvd0/disp: call into core to handle dac power state changes
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 8 Nov 2012 02:01:39 +0000 (12:01 +1000)]
drm/nvd0/disp: call into core to handle sor power state changes
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 7 Nov 2012 06:43:00 +0000 (16:43 +1000)]
drm/nvd0/disp: move link training helpers into core as display methods
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 7 Nov 2012 06:28:35 +0000 (16:28 +1000)]
drm/nouveau/core: allow representing method ranges in nouveau_omthds
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 7 Nov 2012 06:16:41 +0000 (16:16 +1000)]
drm/nouveau/core: expose full method calling capabilities with nv_exec
nv_call() just allows mthd+u32 submission, nv_exec() exposes the
mthd+data+size interface which will be used in future commits.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 6 Nov 2012 06:03:51 +0000 (16:03 +1000)]
drm/nouveau/bios: implement "full" BIT 'd' table (and subtable) parsing in core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 2 Nov 2012 01:33:27 +0000 (11:33 +1000)]
drm/nvd0/disp: move remaining interrupt handling to core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 1 Nov 2012 01:09:53 +0000 (11:09 +1000)]
drm/nouveau/core: add some missing subdev/engine disable flags
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 2 Nov 2012 01:02:47 +0000 (11:02 +1000)]
drm/nouveau/bios: implement BIT 'U' table (and subtable) parsing in core
This will, in the near future, replace what's currently in the DRM
nouveau_bios.c code.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 16 Oct 2012 04:18:32 +0000 (14:18 +1000)]
drm/nvd0/disp: allocate display and evo channels from driver core
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 30 Oct 2012 00:59:12 +0000 (10:59 +1000)]
drm/nouveau: run mode_config destructor before destroying internal display state
Later changes will depend on being able to pull down CRTCs etc with the
master display state still intact.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Mon, 29 Oct 2012 03:36:45 +0000 (13:36 +1000)]
drm/nouveau/drm/nvd0/disp: display->disp
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 16 Oct 2012 04:00:31 +0000 (14:00 +1000)]
drm/nvd0/disp: introduce a nvd0_head as a subclass of nouveau_crtc
This will be used instead of storing a heap of per-head data (such as evo
channels) in nvd0_display in some other way.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 28 Aug 2012 04:10:39 +0000 (14:10 +1000)]
drm/nvd0-nve0/disp: initial implementation of evo channel classes
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 30 Oct 2012 00:03:38 +0000 (10:03 +1000)]
drm/nv50/disp: allow PCI_US pushbuf binding
Ben Skeggs [Tue, 14 Aug 2012 04:11:49 +0000 (14:11 +1000)]
drm/nv50/disp: initial implementation of the various channel object classes
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 14 Aug 2012 02:50:14 +0000 (12:50 +1000)]
drm/nv50/disp: enable interrupts and setup memory area
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 14 Aug 2012 01:29:57 +0000 (11:29 +1000)]
drm/nv50/disp: copy caps to evo
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 14 Aug 2012 00:04:04 +0000 (10:04 +1000)]
drm/nv50/disp: create skeleton display/channel object classes
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Mon, 8 Oct 2012 04:44:00 +0000 (14:44 +1000)]
drm/nvd0/dmaobj: initial bind() method implementation
Currently unused, and rudimentary. Lots to figure out here still, but
this is sufficient for what disp will need.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Mon, 8 Oct 2012 04:34:35 +0000 (14:34 +1000)]
drm/nvc0/dmaobj: implement initial bind() method
Currently unused.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Mon, 8 Oct 2012 04:29:16 +0000 (14:29 +1000)]
drm/nv50/dmaobj: extend class to allow gpu-specific attributes to be defined
disp is going to need to be able to create more specific dma objects
than was previously possible.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Mon, 8 Oct 2012 04:11:35 +0000 (14:11 +1000)]
drm/nvd0/dmaobj: duplicate fermi class, will diverge real soon now
The hardware dmaobj format completely changed in GF119, so these will
need a separate implementation.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Mon, 8 Oct 2012 03:39:13 +0000 (13:39 +1000)]
drm/nvc0/dmaobj: stub bind function for now so we can call unconditionally
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Mon, 8 Oct 2012 02:58:23 +0000 (12:58 +1000)]
drm/nouveau/dmaobj: move parent class check to bind() method
Otherwise when nvc0- gains a bind() method (disp needs it), the fifo
engine will attempt to create a dma object for the push buffer, which
is unnecessary on fermi.
The only sane place to put these checks is in the bind method itself,
and have it unconditionally called from wherever it might be needed.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 29 Aug 2012 00:54:49 +0000 (10:54 +1000)]
drm/nouveau/dmaobj: merge everything except ctor and bind together
Simplifies things a little, and currently no reason to need
chipset-specific dmaobj constructors.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Sat, 17 Nov 2012 11:51:30 +0000 (21:51 +1000)]
drm/nouveau: pass address to object accessor functions as u64
Will be required by future work. Make the API change now to catch any
(but hopefully none) unexpected fallout.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 21 Nov 2012 05:04:51 +0000 (15:04 +1000)]
drm/nouveau/gpio: place upper limit on using old-school tvdac bios data
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 21 Nov 2012 05:02:47 +0000 (15:02 +1000)]
drm/nouveau: remove newline-only NV_DEBUG calls
This used to output the function name, now doesn't, so just looks
stupid.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Marcin Slusarz [Sun, 11 Nov 2012 18:52:59 +0000 (19:52 +0100)]
drm/nv50: decode PGRAPH status registers on TLB flush fail
SIgned-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Viktor Novotný [Sat, 10 Nov 2012 18:24:06 +0000 (19:24 +0100)]
drm/nouveau/clk: fix crystal frequency retrieval on nv25
Signed-off-by: Viktor Novotný <noviktor@seznam.cz>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Marcin Slusarz [Sun, 11 Nov 2012 18:57:35 +0000 (19:57 +0100)]
drm/nouveau: remove unused variable from nouveau_bios_shadow_of
drivers/gpu/drm/nouveau/core/subdev/bios/base.c: warning: unused variable 'i'
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 30 Oct 2012 03:55:13 +0000 (13:55 +1000)]
drm/nv40/fb: use an actual compressed zeta format
SPLIT is apparently just that, and only splits Z and S components.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 11 Oct 2012 05:46:30 +0000 (15:46 +1000)]
drm/nv30/fb: enable z compression
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 11 Oct 2012 05:13:10 +0000 (15:13 +1000)]
drm/nv40/fb: enable z compression
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 12 Oct 2012 01:24:39 +0000 (11:24 +1000)]
drm/nv20/fb: fixup compression tag allocation size
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 11 Oct 2012 04:08:16 +0000 (14:08 +1000)]
drm/nv30-nv40/graph: poke zcomp regs from tile_prog hook
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 10 Oct 2012 02:52:00 +0000 (12:52 +1000)]
drm/nv30-nv40/fb: call zcomp setup hook from tiling setup
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 10 Oct 2012 01:31:11 +0000 (11:31 +1000)]
drm/nouveau/fb: create tag heap from common code for all relevant chipsets
A nv2x bug wrt hardcoded tag counts is now also fixed as a side-effect.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 10 Oct 2012 01:18:00 +0000 (11:18 +1000)]
drm/nv30/fb: start bashing zcomp registers with 'disabled' (for now)
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 10 Oct 2012 02:13:38 +0000 (12:13 +1000)]
drm/nouveau: replace some open-coded mm_initialised checks
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 10 Oct 2012 01:09:48 +0000 (11:09 +1000)]
drm/nouveau/fb: read TILE_BASE after writing it to avoid a hardware race
Apparently needed for turbocache nv4x chips at least, we'll just do it
everywhere...
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 10 Oct 2012 00:57:49 +0000 (10:57 +1000)]
drm/nv40/fb: start bashing zcomp registers on relevant chipsets
Always bashing "disabled" for now, actual compressing coming up...
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 10 Oct 2012 00:41:04 +0000 (10:41 +1000)]
drm/nv30-nv40/fb: make use of bankoff for zeta buffers, where supported
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Wed, 10 Oct 2012 00:34:29 +0000 (10:34 +1000)]
drm/nv20/fb: fix zcomp register calculation on big-endian systems
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 9 Oct 2012 06:03:23 +0000 (16:03 +1000)]
drm/nv40/fb: split implementation into nv40/nv41/nv44/nv46/nv47/nv49/nv4e pieces
Wow, this is a nice complicated mess of build-your-own-mc blocks...
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 9 Oct 2012 05:56:16 +0000 (15:56 +1000)]
drm/nv30/fb: split implementation into nv34(nv10)/nv30/nv35 pieces
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 9 Oct 2012 05:48:11 +0000 (15:48 +1000)]
drm/nv20/fb: split implementation into nv20/nv25 pieces
There's more stuff that can be shared in the constructor, will be merged
together again later.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 9 Oct 2012 05:41:09 +0000 (15:41 +1000)]
drm/nouveau/fb: split tile and compression region handling
This is in preparation for extending the support to the remaining
chipsets, to allow for sharing more functions.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Fri, 12 Oct 2012 02:45:15 +0000 (22:45 -0400)]
drm/nouveau/bios: attempt to fetch entire acpi rom image in one shot
v2: fdo#55948 - the _ROM method silently truncates size to 4KiB, perform
a checksum test and fall back to slow _ROM access on failure.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Daniel Vetter [Thu, 1 Nov 2012 13:45:18 +0000 (14:45 +0100)]
drm/doc: add new dp helpers into drm DocBook
I didn't bother with documenting the really trivial new "extract
something from dpcd" helpers, but the i2c over aux ch is now
documented a bit.
v2: Clarify the comment for i2c_dp_aux_add_bus a bit.
v3: Fix more spelling fail spotted by Laurent Pinchart.
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Daniel Vetter [Thu, 1 Nov 2012 13:45:17 +0000 (14:45 +0100)]
drm/doc: integrate fb helper reference into docs
Again only minimal changes to make kerneldoc no longer shout. Plus a
little introduction in the form of a inline DOC: section to quickly
explain what this is all about.
v2: Fixup spelling fail.
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Daniel Vetter [Thu, 1 Nov 2012 13:45:16 +0000 (14:45 +0100)]
drm/doc: integrate crtc helper api into docbook
- Add the missing doc for drm_helper_move_panel_connectors_to_head.
- Fixup any outdated stuff in existing sections. I've only looked at
those kerneldoc headers that actually resulted in a complaint from
the kerneldoc parser tool.
v2:
- Actually include the docbook snippet in the right patch.
- Fix spelling fail.
v3: It's now called drm_crtc_helper_set_mode, spotted by Laurent
Pinchart.
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Daniel Vetter [Thu, 1 Nov 2012 13:45:15 +0000 (14:45 +0100)]
drm/doc: Helpers are not a Midlayer!
I'm devoting all my wrath to that fight, so don't misname it ;-)
v2: Make it clear that this section talks about kms helpers.
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Daniel Vetter [Thu, 22 Nov 2012 08:53:55 +0000 (09:53 +0100)]
drm/edid: tune down debug message in parse_hdmi_vsdb
Those tend to be totally not interesting for end-users, and for
debugging we tend to dump the entire noise anyway by enabling all
debug messages.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57388
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
James Hogan [Fri, 23 Nov 2012 11:12:10 +0000 (11:12 +0000)]
drm: only build ati_pcigart if PCI enabled
Prevent ati_pcigart.c being built unless PCI is enabled. The exported
functions in this file are only used by drivers which depend on PCI
(namely r128 and radeon), and it tries to use PCI specific functions
(pci_unmap_page, pci_map_page, and pci_dma_mapping_error) that cause
compiler errors when PCI is disabled.
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Wed, 21 Nov 2012 15:04:18 +0000 (16:04 +0100)]
drm/vmwgfx: Tighten the security around buffer maps
Make sure that other DRM clients can't map the contents of
non-shareable buffer objects.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thierry Reding [Wed, 21 Nov 2012 08:50:41 +0000 (09:50 +0100)]
drm: tegra: Add Tegra30 support
Add support for host1x, the display controllers and HDMI on the Tegra30
SoC.
Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Mark Zhang <markz@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thierry Reding [Thu, 22 Nov 2012 19:37:17 +0000 (19:37 +0000)]
drm: tegra: Use framebuffer pitch as line stride
Instead of using the stride derived from the display mode, use the pitch
associated with the currently active framebuffer. This fixes a bug where
the LCD display content would be skewed when enabling HDMI with a video
mode different from that of the LCD.
Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Tested-by: Terje Bergstrom <tbergstrom@nvidia.com>
Tested-by: Mark Zhang <markz@nvidia.com>
Tested-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Wed, 21 Nov 2012 14:53:21 +0000 (14:53 +0000)]
drm/ttm: Optimize vm locking using kref_get_unless_zero v3
Removes the need for a write lock each time we call ttm_bo_unref().
v2: Remove an unused variable.
v3: Really remove the unused variable.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 20 Nov 2012 12:16:51 +0000 (12:16 +0000)]
drm/ttm: Use the hashtab _rcu interface for ttm_objects
Also move a kref_init() out of spinlocked region
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 20 Nov 2012 12:16:50 +0000 (12:16 +0000)]
drm/ttm: Fix locking in an error path
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 20 Nov 2012 12:16:49 +0000 (12:16 +0000)]
drm/vmwgfx: Free user-space fence objects correctly
They need to be freed after an rcu grace period.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 20 Nov 2012 12:16:48 +0000 (12:16 +0000)]
kref: Add kref_get_unless_zero documentation
Document how kref_get_unless_zero should be used and how it helps
solve a typical kref / locking problem.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 20 Nov 2012 12:16:47 +0000 (12:16 +0000)]
drm: Add a hash-tab rcu-safe API
While hashtab should now be RCU-safe, Add a drm_ht_xxx_api for consumers
to use to make it obvious what locking mechanism is used.
Document the way the rcu-safe interface should be used.
Don't use rcu-safe list traversal in modify operations where we should use
a spinlock / mutex anyway.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 20 Nov 2012 12:19:37 +0000 (12:19 +0000)]
drm/vmwgfx: Add and make use of a header for surface size calculation.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 20 Nov 2012 12:19:36 +0000 (12:19 +0000)]
drm/vmwgfx: Break out surface and context management to separate files
Add a resource-private header for common resource definitions
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 20 Nov 2012 12:19:35 +0000 (12:19 +0000)]
drm/vmwgfx: Refactor resource management
Refactor resource management to make it easy to hook up resources
that are backed up by buffers. In particular, resources and their
backing buffers can be evicted and rebound, if supported by the device.
To avoid query deadlocks, the query code is also modified somewhat.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Fri, 9 Nov 2012 12:26:15 +0000 (12:26 +0000)]
drm/vmwgfx: Make vmw_dmabuf_unreference handle NULL objects
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Fri, 9 Nov 2012 12:26:14 +0000 (12:26 +0000)]
drm/vmwgfx: Refactor module load to not require fifo unless fbdev is loaded
This also fixes a bug where the fence manager was left without irq
enabled when waiting for fences, causing various errors at module
load time
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Fri, 9 Nov 2012 12:26:13 +0000 (12:26 +0000)]
drm/vmwgfx: Make screen object code not require fifo at init time
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Fri, 9 Nov 2012 12:26:12 +0000 (12:26 +0000)]
drm/vmwgfx: Make overlay code not require fifo at init time
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Fri, 9 Nov 2012 12:26:11 +0000 (12:26 +0000)]
drm/vmwgfx: Enable traces *after* we've hidden SVGA
Hiding SVGA seems to trigger a VGA screen clear, and with no
traces dirty it doesn't seem to repaint
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Maarten Lankhorst [Tue, 6 Nov 2012 13:39:43 +0000 (14:39 +0100)]
drm/ttm: alter cpu_writers to return -EBUSY in ttm_execbuf_util reservations
This is similar to other platforms that don't allow command submission
to buffers locked on the cpu.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 6 Nov 2012 11:31:51 +0000 (11:31 +0000)]
drm/ttm: Optimize reservation slightly
Reservation locking currently always takes place under the LRU spinlock.
Hence, strictly there is no need for an atomic_cmpxchg call; we can use
atomic_read followed by atomic_write since nobody else will ever reserve
without the lru spinlock held.
At least on Intel this should remove a locked bus cycle on successful
reserve.
Note that thit commit may be obsoleted by the cross-device reservation work.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 6 Nov 2012 11:31:50 +0000 (11:31 +0000)]
drm/ttm, drm/vmwgfx: Use RCU locking for object lookups v3
The mostly used lookup+get put+potential_destroy path of TTM objects
is converted to use RCU locks. This will substantially decrease the amount
of locked bus cycles during normal operation.
Since we use kfree_rcu to free the objects, no rcu synchronization is needed
at module unload time.
v2: Don't touch include/linux/kref.h
v3: Adapt to kref_get_unless_zero return value change
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 6 Nov 2012 11:31:49 +0000 (11:31 +0000)]
kref: Implement kref_get_unless_zero v3
This function is intended to simplify locking around refcounting for
objects that can be looked up from a lookup structure, and which are
removed from that lookup structure in the object destructor.
Operations on such objects require at least a read lock around
lookup + kref_get, and a write lock around kref_put + remove from lookup
structure. Furthermore, RCU implementations become extremely tricky.
With a lookup followed by a kref_get_unless_zero *with return value check*
locking in the kref_put path can be deferred to the actual removal from
the lookup structure and RCU lookups become trivial.
v2: Formatting fixes.
v3: Invert the return value.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thomas Hellstrom [Tue, 6 Nov 2012 11:31:48 +0000 (11:31 +0000)]
drm: Make hashtab rcu-safe
TTM base objects will be the first consumer.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Maarten Lankhorst [Fri, 12 Oct 2012 15:04:00 +0000 (15:04 +0000)]
drm/ttm: remove sync_arg from driver functions
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Reviewed-By: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Maarten Lankhorst [Fri, 12 Oct 2012 15:03:11 +0000 (15:03 +0000)]
drm/ttm: remove sync_obj_arg from ttm_bo_move_accel_cleanup
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Reviewed-By: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Maarten Lankhorst [Fri, 12 Oct 2012 15:02:19 +0000 (15:02 +0000)]
drm/ttm: remove sync_obj_arg member
vmwgfx was its only user and always sets it to the same..
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Reviewed-By: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Maarten Lankhorst [Fri, 12 Oct 2012 15:01:43 +0000 (15:01 +0000)]
drm/vmwgfx: remove use of fence_obj_args
It's always hardcoded to the same value.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Reviewed-By: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Egbert Eich [Sat, 13 Oct 2012 11:36:14 +0000 (11:36 +0000)]
DRM/KMS: Add Bail-Out Conditions for Loop.
When trying to obtain an accurate timestamp for the last vsync interrupt
in vblank_disable_and_save() we loop until the vsync counter after reading
the time stamp is identical to the one before.
In the case where no hardware timestamp can be obtained there is probably
no point in trying to make sure we remain within the same vsync during
the time we obtain the counter.
Furthermore we should make sure there's an 'emergency exit' so that we
don't end up in an endless loop when the driver get_vblank_timestamp()
function doesn't manage to return within the same vsync.
This may happen when this function prints out debugging information over
a slow (ie serial) line.
Signed-off-by: Egbert Eich <eich@suse.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Daniel Vetter [Wed, 24 Oct 2012 13:35:50 +0000 (13:35 +0000)]
drm: don't unnecessarily enable the polling work
... by properly checking connector->polled. This doesn't matter too
much because the polling work itself gets this slightly more right and
doesn't set repoll if there's nothing to do. But we can do better.
v2: Chris Wilson noticed that I broke polling, since repoll will never
ever be set true. Fix this up, and simplify the logic a bit while at
it.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
projects / GitHub / exynos8895 / android_kernel_samsung_universal8895.git / log