openwrt/staging/blogic.git
7 years agodrm/ttm: add TTM_PL_FLAG_CONTIGUOUS v2
Christian König [Wed, 29 Mar 2017 10:13:54 +0000 (12:13 +0200)]
drm/ttm: add TTM_PL_FLAG_CONTIGUOUS v2

This allows drivers to specify if they need a contiguous allocation or not.

v2: use space instead of tab

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/ttm: add io_mem_pfn callback
Christian König [Tue, 28 Mar 2017 14:54:50 +0000 (16:54 +0200)]
drm/ttm: add io_mem_pfn callback

This allows the driver to handle io_mem mappings on their own.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/ttm: cleanup and optimize ttm_bo_mem_compat v2
Christian König [Wed, 29 Mar 2017 09:47:04 +0000 (11:47 +0200)]
drm/ttm: cleanup and optimize ttm_bo_mem_compat v2

No need to implement the same logic twice. Also check if the busy placements
are identical to the already scanned placements before checking them.

v2: improve check even more as suggested by Michel.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu/vi: add defines for KIQ packets
Alex Deucher [Thu, 30 Mar 2017 16:24:59 +0000 (12:24 -0400)]
drm/amdgpu/vi: add defines for KIQ packets

Make it clearer how these packets work.

Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu/soc15: Fix static checker warnings
Alex Deucher [Mon, 3 Apr 2017 20:56:08 +0000 (16:56 -0400)]
drm/amdgpu/soc15: Fix static checker warnings

vega10 is the only soc15 asic at the moment so these
warnings are invalid, but add a default case to silence
the warnings.

Fixes: 220ab9bd1ccf: "drm/amdgpu: soc15 enable (v3)"
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Read vram width from integrated system info table
Harry Wentland [Sat, 1 Apr 2017 00:14:33 +0000 (20:14 -0400)]
drm/amdgpu: Read vram width from integrated system info table

On KB, KV, CZ we should read the vram width from integrated system
table, if we can. The NOOFCHAN in MC_SHARED_CHMAP is not accurate.

With this change we can enable two 4k displays on CZ again. This use
case was broken sometime in January when we started looking at
vram_width for bandwidth calculations instead of hardcoding this value.

v2:
  Return 0 if integrated system info table is not available.

Tested-by: Roman Li <roman.li@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: create a func to check vm size
Zhang, Jerry [Wed, 29 Mar 2017 08:08:31 +0000 (16:08 +0800)]
drm/amdgpu: create a func to check vm size

break it out from the check parameters function.

Signed-off-by: Junwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Fix a NULL deref in amdgpu_vm_add_prt_cb()
Dan Carpenter [Mon, 3 Apr 2017 18:41:39 +0000 (21:41 +0300)]
drm/amdgpu: Fix a NULL deref in amdgpu_vm_add_prt_cb()

We accidentally dereference "cb" if the kmalloc() fails.

Fixes: 451bc8eb8fe6 ("drm/amdgpu: fix PRT teardown on VM fini v3")
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: fix a couple locking issues
Dan Carpenter [Mon, 3 Apr 2017 18:42:42 +0000 (21:42 +0300)]
drm/amd/powerplay: fix a couple locking issues

We should return unlock on the error path in pp_dpm_dispatch_tasks()
and there is a double lock bug in pp_dpm_set_sclk_od().

Fixes: 2a5071056e6a ("drm/amd/powerplay: add global PowerPlay mutex.")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/powerplay: fix pp_dpm_get_current_power_state() (v2)
Dan Carpenter [Mon, 3 Apr 2017 18:41:47 +0000 (21:41 +0300)]
drm/amd/powerplay: fix pp_dpm_get_current_power_state() (v2)

This switch statement is missing breaks.

v2: agd: break in default case as well

Fixes: 2a5071056e6a ("drm/amd/powerplay: add global PowerPlay mutex.")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: various cleanups for uvd/vce.
Rex Zhu [Thu, 30 Mar 2017 07:45:52 +0000 (15:45 +0800)]
drm/amdgpu: various cleanups for uvd/vce.

Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: when resume failed, return error to avoid system hang.
Rex Zhu [Thu, 30 Mar 2017 05:21:01 +0000 (13:21 +0800)]
drm/amdgpu: when resume failed, return error to avoid system hang.

Continuing if the GPU fails to resume will end in pain.

Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: use a 64bit interval tree for VM management v2
Christian König [Thu, 30 Mar 2017 12:03:59 +0000 (14:03 +0200)]
drm/amdgpu: use a 64bit interval tree for VM management v2

This only makes a difference for 32-bit systems. The idea is to have a
fixed virtual address space size with 4-level page tables and to
minimize differences between 32 and 64-bit systems.

v2: Update commit message.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: fix semicolon.cocci warnings
kbuild test robot [Fri, 31 Mar 2017 10:15:10 +0000 (18:15 +0800)]
drm/amdgpu: fix semicolon.cocci warnings

drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c:133:2-3: Unneeded semicolon

 Remove unneeded semicolon.

Generated by: scripts/coccinelle/misc/semicolon.cocci

Acked-by: Huang Rui <ray.huang@amd.com>
CC: Huang Rui <ray.huang@amd.com>
Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu/powerplay: fix locking typo
Alex Deucher [Fri, 31 Mar 2017 14:51:29 +0000 (10:51 -0400)]
drm/amdgpu/powerplay: fix locking typo

Fixes: 2a5071056e6a601e ("drm/amd/powerplay: add global PowerPlay mutex.")
Reported-by: Julia Lawall <julia.lawall@lip6.fr>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agoMerge tag 'omapdrm-4.12' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux...
Dave Airlie [Mon, 3 Apr 2017 19:45:49 +0000 (05:45 +1000)]
Merge tag 'omapdrm-4.12' of git://git./linux/kernel/git/tomba/linux into drm-next

omapdrm changes for 4.12

Main changes include:

* Add support for render nodes.
* Refactor omapdss code to allow multiple DISPC implementations. This is pre-work for DSS6 support.
* Fix replication logic bug, which caused RGB565 fb to be shown too dark on a 24bit display.
* Improve detection of display stack readiness, which should remove the probe order issues.
* Link panel-dpi with its backlight, so that they are turned on/off in sync.
* Fix possibly incorrect setup of sync and data-enable signals.
* Get rid of DRM_OMAP_NUM_CRTCS config option.

* tag 'omapdrm-4.12' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux: (34 commits)
  drm/omap: Major omap_modeset_init() cleanup
  drm/omap: Remove the obsolete #define omap_plane _omap_plane hack
  drm/omap: Fix one ugly indentation style break left by coccinelle
  drm/omap: Rename enum omap_plane to enum omap_plane_id
  drm/omap: Get rid of DRM_OMAP_NUM_CRTCS config option
  drm/omap: fix crash on module unload
  drm/omap: use drm_atomic_helper_shutdown()
  drm/omap: fix display SYNC/DE flags
  drm/omap: dispc: improve debug print of display flags
  drm/omap: displays: panel-dpi: Support for handling backlight devices
  drm/omap: poll only connectors where the connect/disconnect can be checked
  drm/omap: display: Add displays in sorted order to the panel_list
  drm/omap: Use omapdss_stack_is_ready() to check that the display stack is up
  drm/omap: dss: Support for detecting display stack readiness
  drm/omap: dss: Functions to check components in the display/output list
  drm/omap: fix replication logic
  drm/omap: remove unused dispc_wb_enable & dispc_wb_is_enabled
  drm/omap: remove all EXPORT_SYMBOLs from dispc.c
  drm/omap: use dispc_ops
  drm/omap: fill dispc_ops
  ...

7 years agodrm/omap: Major omap_modeset_init() cleanup
Jyri Sarha [Fri, 24 Mar 2017 14:47:55 +0000 (16:47 +0200)]
drm/omap: Major omap_modeset_init() cleanup

Cleanup overly complex omap_modeset_init(). The function is trying to
support many unusual configuration, that have never been tested and
are not supported by other parts of the dirver.

After cleanup the init function creates exactly one connector,
encoder, crtc, and primary plane per each connected dss-device. Each
connector->encoder->crtc chain is expected to be separate and each
crtc is connect to a single dss-channel. If the configuration does not
match the expectations or exceeds the available resources, the
configuration is rejected.

Signed-off-by: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: Remove the obsolete #define omap_plane _omap_plane hack
Jyri Sarha [Fri, 24 Mar 2017 14:47:54 +0000 (16:47 +0200)]
drm/omap: Remove the obsolete #define omap_plane _omap_plane hack

Remove the obsolete "#define omap_plane _omap_plane" hack and other
related hacks to get around the enum omap_plane colliding with struct
omap_plane.

Signed-off-by: Jyri Sarha <jsarha@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: Fix one ugly indentation style break left by coccinelle
Jyri Sarha [Fri, 24 Mar 2017 14:47:53 +0000 (16:47 +0200)]
drm/omap: Fix one ugly indentation style break left by coccinelle

Fix one ugly indentation style break left by the previous coccilnelle
patch.

Signed-off-by: Jyri Sarha <jsarha@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: Rename enum omap_plane to enum omap_plane_id
Jyri Sarha [Fri, 24 Mar 2017 14:47:52 +0000 (16:47 +0200)]
drm/omap: Rename enum omap_plane to enum omap_plane_id

The enum omap_plane conflicted with the same struct name for omapdrm
plane private data. This rename should solve the conflict.

The rename was implement with this very simple coccinelle patch:
------------------------
@@
@@
enum
-omap_plane
+omap_plane_id
------------------------
The patch was applied like this:
spatch --sp-file <cocci_file> --all-includes --in-place --dir drivers/gpu/drm/omapdrm

The above patch did not rename the actual enum definition. That was
added manually on top of the spatch changes.

Signed-off-by: Jyri Sarha <jsarha@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: Get rid of DRM_OMAP_NUM_CRTCS config option
Jyri Sarha [Fri, 24 Mar 2017 14:47:51 +0000 (16:47 +0200)]
drm/omap: Get rid of DRM_OMAP_NUM_CRTCS config option

Allocate one CRTC for each connected output and get rid of
DRM_OMAP_NUM_CRTCS config option. We still can not create more CRTCs
than we have DSS display managers. We also reserve one overlay per
CRTC for primary plane so we can not have more CRTCs than we have
overlays either.

Signed-off-by: Jyri Sarha <jsarha@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: fix crash on module unload
Tomi Valkeinen [Fri, 24 Mar 2017 07:58:49 +0000 (09:58 +0200)]
drm/omap: fix crash on module unload

When unloading omapdrm we get a NULL pointer deref in
omap_drm_irq_uninstall(). This is caused by:

967dd48417874dd25491a4e933648f394a64f70f ("drm: remove
drm_vblank_no_hw_counter assignment from driver code")

We shut down all the crtcs at unload time before calling
omap_drm_irq_uninstall, so the code in omap_drm_irq_uninstall() is
unnecessary and can be removed.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agodrm/omap: use drm_atomic_helper_shutdown()
Tomi Valkeinen [Mon, 27 Mar 2017 07:02:22 +0000 (10:02 +0300)]
drm/omap: use drm_atomic_helper_shutdown()

Use drm_atomic_helper_shutdown() to ensure that all crtcs are disabled
when unloading the driver.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: fix display SYNC/DE flags
Tomi Valkeinen [Wed, 23 Nov 2016 11:24:00 +0000 (13:24 +0200)]
drm/omap: fix display SYNC/DE flags

At the moment VSYNC/HSYNC/DE high/low flags set by the panel/encoder
drivers get lost when the videotimings are translated to DRM's
videomode, as DRM's mode does not have corresponding flags.

DRM has bus-flags for this purpose, and while it lacks a few flags at
the moment, it should be used here. However, until we rewrite omapdrm's
device model, using bus-flags is rather difficult.

As a short term fix, this patch makes sure that every time the videomode
is set in omap_crtc_mode_set_nofb(), the driver asks for the SYNC/DE
flags from the panel/encoder drivers, and thus we get the correct flags
into use.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: dispc: improve debug print of display flags
Tomi Valkeinen [Wed, 23 Nov 2016 11:23:42 +0000 (13:23 +0200)]
drm/omap: dispc: improve debug print of display flags

Instead of printing 0/1 for display flags like vsync high/low, use a
tri-state print (-1/0/1) to indicate the "undefined" state.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: displays: panel-dpi: Support for handling backlight devices
Peter Ujfalusi [Fri, 9 Sep 2016 10:33:06 +0000 (13:33 +0300)]
drm/omap: displays: panel-dpi: Support for handling backlight devices

The associated backlight device can be configured via DT by providing the
phandle to the device.

If the backlight device is configured, the driver can manage the backligt
along with the panel's power state, iow it can turn on the backlight when
the panel is enabled and turn it off when the panel is disabled.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: poll only connectors where the connect/disconnect can be checked
Peter Ujfalusi [Mon, 22 Aug 2016 12:54:30 +0000 (15:54 +0300)]
drm/omap: poll only connectors where the connect/disconnect can be checked

When the connector associated detect callback is not provided, we can not
detect if the display is connected or disconnected. These displays do not
support hot plug, they are always connected. Let DRM know that connectors
w/o detect callback should not be polled.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agodrm/omap: display: Add displays in sorted order to the panel_list
Peter Ujfalusi [Thu, 11 Aug 2016 07:20:23 +0000 (10:20 +0300)]
drm/omap: display: Add displays in sorted order to the panel_list

Keep the panel_list ordered according to aliases. The DRM connectors will
be created following the panel_list. By keeping the list ordered the DRM
connectors will be created in the same order regardless of the driver
probe order.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: Use omapdss_stack_is_ready() to check that the display stack is up
Peter Ujfalusi [Tue, 3 May 2016 19:08:01 +0000 (22:08 +0300)]
drm/omap: Use omapdss_stack_is_ready() to check that the display stack is up

Instead of 'guessing' based on aliases of the status of the DSS drivers,
use the new interface to check that all needed drivers are loaded.
In this way we can be sure that all needed drivers are loaded so it is
safe to continue the probing of omapdrm.
This method will allow the omapdrm to be probed 'headless', without
outputs.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: dss: Support for detecting display stack readiness
Peter Ujfalusi [Tue, 3 May 2016 19:07:10 +0000 (22:07 +0300)]
drm/omap: dss: Support for detecting display stack readiness

When omapdss is loaded (all core components are in place) create a list of
devices used in the display graph. This list later can be used by omapdrm
via the omapdss_stack_is_ready() function to check that these components
are loaded. Based on this information, omapdrm can defer probe in case when
the omapdss stack is not ready yet.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: dss: Functions to check components in the display/output list
Peter Ujfalusi [Mon, 2 May 2016 11:55:38 +0000 (14:55 +0300)]
drm/omap: dss: Functions to check components in the display/output list

The functions can be used to check a component (by it's of_node) if it is
part of the omapdss display or output list. If the component is found, it
means that the driver is loaded.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: fix replication logic
Tomi Valkeinen [Mon, 29 Aug 2016 10:15:02 +0000 (13:15 +0300)]
drm/omap: fix replication logic

DSS uses "replication logic" to convert color components from smaller
bit widths to bigger bit widths. Without replication logic, the color
component would be shifted and the least significant bits would be left
at 0, whereas with replication logic, the least significat bits will be
filled with the most significant bit.

For some reason omapdrm leaves replication logic always to off, which,
in the most common case, means that showing 16-bit RGB colors on 24-bit
display gives a bit darker colors than expected.

The reset value of the replication logic bit is enabled, and it should
always be set, as there's no practical reason to disable it. This patch
removes the option to disable it from omapdrm, and always sets it to
enabled in dispc.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agodrm/omap: remove unused dispc_wb_enable & dispc_wb_is_enabled
Tomi Valkeinen [Tue, 10 Nov 2015 23:59:49 +0000 (17:59 -0600)]
drm/omap: remove unused dispc_wb_enable & dispc_wb_is_enabled

Remove two unused WB functions.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agodrm/omap: remove all EXPORT_SYMBOLs from dispc.c
Tomi Valkeinen [Thu, 5 Nov 2015 18:06:06 +0000 (20:06 +0200)]
drm/omap: remove all EXPORT_SYMBOLs from dispc.c

omapdrm now uses dispc_ops instead of direct function calls so we can
remove all EXPORT_SYMBOLs from dispc. Most of the functions can also be
made static, but a few are used outside dispc.c.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: use dispc_ops
Tomi Valkeinen [Thu, 5 Nov 2015 16:39:52 +0000 (18:39 +0200)]
drm/omap: use dispc_ops

Change omapdrm to get dispc_ops and use that to call the dispc functions
instead or direct function calls.

The change is very straightforward.

The only problem was in omap_crtc_init() which calls pipe2vbl(crtc), and
at that point of time the crtc->dev link, which is used to get the
dispc_ops, has not been set up yet. This patch makes omap_crtc_init()
skip the call to pipe2vbl() and instead calls
dispc_ops->mgr_get_vsync_irq() directly.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: fill dispc_ops
Tomi Valkeinen [Thu, 5 Nov 2015 17:44:38 +0000 (19:44 +0200)]
drm/omap: fill dispc_ops

This patch changes the current omapdss driver to fill a dispc_ops struct
and set it to omapdss-base.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: add dispc_ops
Tomi Valkeinen [Thu, 5 Nov 2015 17:36:02 +0000 (19:36 +0200)]
drm/omap: add dispc_ops

We want to change the dispc API from plain functions to a struct with
functions pointers, so that omapdrm can call either omapdss or omapdss6
depending on the platform.

This patch adds 'struct dispc_ops' and adds functions to omapdss-base
to set and get the ops.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: move dispc related dss-feat funcs to dispc
Tomi Valkeinen [Thu, 5 Nov 2015 17:54:33 +0000 (19:54 +0200)]
drm/omap: move dispc related dss-feat funcs to dispc

omapdrm still uses a few non-dispc functions: dss_feat_get_num_mgrs(),
dss_feat_get_num_ovls() and  dss_feat_get_supported_color_modes(). We
want to provide omapdrm a single dispc_ops function pointer struct so
that omapdrm will use either the current omapdss or the new omapdss6
driver depending on the platform.

Those three functions are really dispc functions, but are located in the
dss_features.c for legacy reasons.

This patch adds similar functionss to the dispc, and changes omapdrm to
use those new functions. Underneath the functions still call the
functions from dss_feature.c.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: move display, dss-of, output to omapdss-base
Tomi Valkeinen [Thu, 5 Nov 2015 15:26:18 +0000 (17:26 +0200)]
drm/omap: move display, dss-of, output to omapdss-base

This patch moves the common parts of omapdss to omapdss-base so that
both the current omapdss driver and the new omapdss6 driver can use
them.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: display: don't use dsi_get_pixel_size()
Tomi Valkeinen [Tue, 10 Nov 2015 13:50:53 +0000 (15:50 +0200)]
drm/omap: display: don't use dsi_get_pixel_size()

display.c uses dsi_get_pixel_size() which is implemented in the DSI
driver, and we won't have that in the omapdss-base module, to which we
want to move display.c

This patch changes display.c not to use dsi_get_pixel_size(). The call
can be replaced with a simple check for OMAP_DSS_DSI_FMT_RGB565.

We can also make dsi_get_pixel_size() static as it's no longer used
outside dsi.c.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: output: use dev_err instead of DSSERR
Tomi Valkeinen [Fri, 6 Nov 2015 13:26:39 +0000 (15:26 +0200)]
drm/omap: output: use dev_err instead of DSSERR

We don't have omapdss's custom error printing functions in the common
omapdss-base module, to which we want to move output.c.

This patch changes output.c to use dev_err instead of DSSERR so that it
doesn't depend on DSSERR.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agodrm/omap: move dss_initialized to omapdss-base
Tomi Valkeinen [Thu, 5 Nov 2015 15:23:14 +0000 (17:23 +0200)]
drm/omap: move dss_initialized to omapdss-base

omapdss_is_initialized() is used to find out if omapdss has been probed
successfully. This patch moves the related code to the common
omapdss-base module, so that the same support will be there for both
omapdss and omapdss6.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: add omapdss-base.ko
Tomi Valkeinen [Thu, 5 Nov 2015 15:19:32 +0000 (17:19 +0200)]
drm/omap: add omapdss-base.ko

We are working towards enabling omapdss6, which will consists of a new
dss, dispc and dpi drivers. omapdss6 will be a new module. The panel,
encoder and omapdrm will need to use either the current omapdss driver
or the new omapdss6 driver, depending on the platform.

This will be implemented with a common base module and function
pointers.

This patch adds a skeleton omapdss-base.ko module, to which we'll be
moving common dss functionality like registration of the panels.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: fix HDMI sync polarities
Tomi Valkeinen [Fri, 27 May 2016 10:49:05 +0000 (13:49 +0300)]
drm/omap: fix HDMI sync polarities

While implementing writeback support, odd behavior of WBDELAYCOUNT was
observed with the combination of WB capture and HDMI. The result of the
debugging was that the HDMI sync polarities are not set correctly.

The current code sets the sync polarities going from HDMI WP to DISPC
according to the video mode used, which seems to work normally fine, but
causes problems with WB as WB expects the syncs to be active-high.

This patch changes the HDMI sync polarities so that the DISPC always
gets active-high syncs from HDMI WP, and the HDMI core gets sync
polarities according to the used video mode.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agodrm/omap: Add support for render nodes
Hemant Hariyani [Tue, 7 Jun 2016 18:23:19 +0000 (13:23 -0500)]
drm/omap: Add support for render nodes

Add support for render nodes in omap driver and allow required
ioctls to be accessible via render nodes.

This enables unprivileged clients to allocate resources like GEM buffers
for rendering their content into. Mode setting (KMS ioctls) is not
allowed using render nodes. These buffers are then shared with
a previleged process (e.g compositor) that has mode setting access.

An example of this use case is Android where the hardware composer is
the only master and has mode setting access. Every other client then
uses render node(e.g /dev/dri/renderD128 to allocate and use its buffers.

Signed-off-by: Hemant Hariyani <hemanthariyani@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: improve DPI clock selection on DRA7xx
Tomi Valkeinen [Wed, 10 Aug 2016 08:04:29 +0000 (11:04 +0300)]
drm/omap: improve DPI clock selection on DRA7xx

The clock source selection for the LCD outputs is too hardcoded at the
moment. For example, LCD3 is set to use PLL2_1, and PLL2 doesn't exist
on DRA72x SoCs.

There are quite many ways to configure the clocks, even using HDMI PLL
for LCD outputs, but enabling full configuration of the clocks is rather
tricky.

This patch improves the situation a bit by checking if the PLL about to
be used exists, and if not, tries another one.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: decrease min width & height
Tomi Valkeinen [Tue, 23 Aug 2016 09:35:39 +0000 (12:35 +0300)]
drm/omap: decrease min width & height

mode_config's min_width and min_height are both set to 32, which is
overly restrictive.

The real limits depend on whether we're configuring a crtc or a plane,
but a limit of 8x2 is safe for both cases.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agodrm/omap: remove divider constraint from hsdiv
Tomi Valkeinen [Mon, 8 Aug 2016 10:40:35 +0000 (13:40 +0300)]
drm/omap: remove divider constraint from hsdiv

The driver only uses even dividers for hsdiv when pclk >= 100MHz, as odd
dividers can create uneven duty cycle. However, while this holds true
for some dividers like DISPC's LCK and PCK dividers, it is not actually
true for hsdiv.

hsdiv always produces even duty cycle, so the constraint can be removed.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agodrm/omap: refactor CRTC HW property setup
Tomi Valkeinen [Mon, 10 Aug 2015 09:08:50 +0000 (12:08 +0300)]
drm/omap: refactor CRTC HW property setup

The current driver doesn't expose any of the CRTC HW properties like
background color or transparency key, and sets them at CRTC enable time.

Refactor this into a separate function and call that function from
omap_crtc_atomic_flush(). This is the behavior we want when the
properties can be configured, so this patch makes it easier to add
patches later which implement those properties.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
7 years agodrm/omap: work-around for errata i886
Tomi Valkeinen [Tue, 5 Jan 2016 09:43:13 +0000 (11:43 +0200)]
drm/omap: work-around for errata i886

DRA7 errata i886 (FPDLink PLL Unlocks With Certain SoC PLL M/N Values)
says that FPDLink is sensitive to jitter on the vout clock, and that low
PLL M and N values result in more jitter than high M and N values.

This patch implements a workaround for the problem by changing the PLL
setup to search for clocks starting from high M and N values, instead of
low values. This should not cause any functional change, and only
reduces the jitter.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
7 years agoMerge tag 'drm-misc-next-2017-03-31' of git://anongit.freedesktop.org/git/drm-misc...
Dave Airlie [Mon, 3 Apr 2017 06:30:24 +0000 (16:30 +1000)]
Merge tag 'drm-misc-next-2017-03-31' of git://anongit.freedesktop.org/git/drm-misc into drm-next

drm-misc for 4.12:

Core:
- Removed some fb subsampling dimension checks from core (Ville)
- Some MST slot cleanup (Dhinakaran)
- Extracted drm_debugfs.h & drm_ioctl.h from drmP.h (Daniel)
- Added drm_atomic_helper_shutdown() to compliment suspend/resume counterparts
  (Daniel)
- Pipe context through legacy modeset to remove legacy_backoff nasties (Daniel)
- Cleanups around vblank as well as allowing lockless counter reads (Chris W.)
- VGA Switcheroo added to MAINTAINERS with Lukas Wunner as reviewer (Lukas)

Drivers:
- Enhancements to rockchip driver probe (Jeffy) and dsi (Chris Z.)
- Thunderbolt external GPU awareness added (Lukas)

* tag 'drm-misc-next-2017-03-31' of git://anongit.freedesktop.org/git/drm-misc: (63 commits)
  apple-gmux: Don't switch external DP port on 2011+ MacBook Pros
  drm/nouveau: Don't register Thunderbolt eGPU with vga_switcheroo
  drm/amdgpu: Don't register Thunderbolt eGPU with vga_switcheroo
  drm/radeon: Don't register Thunderbolt eGPU with vga_switcheroo
  PCI: Recognize Thunderbolt devices
  MAINTAINERS: Add Lukas Wunner as reviewer for vga_switcheroo
  drm: Fix locking gotcha in page_flip ioctl
  drm: Clarify the role of plane_state argument to drm_simple update().
  drm: Clear e after kfree in drm_mode_page_flip_ioctl
  drm: Convert cmpxchg(bool) back to a two step operation
  drm/bridge: ti-tfp410: support hpd via gpio
  drm: use .hword to represent 16-bit numbers
  Revert unrelated part of "drm: simplify the locking in the GETCRTC ioctl"
  drm: Fixup failure paths in drm_atomic_helper_set_config
  drm: Peek at the current counter/timestamp for vblank queries
  drm: Refactor vblank sequence number comparison
  drm: vblank cannot be enabled if dev->irq_enabled is false
  drm: Mark up accesses of vblank->enabled outside of its spinlock
  drm: Make the decision to keep vblank irq enabled earlier
  drm/atomic-helper: Remove the backoff hack from set_config
  ...

7 years agoMerge branch 'drm-vmwgfx-next' of git://people.freedesktop.org/~syeh/repos_linux...
Dave Airlie [Sun, 2 Apr 2017 06:10:55 +0000 (16:10 +1000)]
Merge branch 'drm-vmwgfx-next' of git://people.freedesktop.org/~syeh/repos_linux into drm-next

This series enables atomic mode set for vmwgfx.  A number of features and
larger fixes are also included.

* 'drm-vmwgfx-next' of git://people.freedesktop.org/~syeh/repos_linux: (22 commits)
  drm/vmwgfx: Properly check display/scanout surface size
  drm/vmwgfx: Support topology greater than texture size
  drm/vmwgfx: Define an overlaid handle_close ioctl.
  drm/vmwgfx: Re-implement the stream resource as a simple resource.
  drm/vmwgfx: Introduce a simple resource type
  drm/vmwgfx: Revert "drm/vmwgfx: Replace numeric parameter like 0444 with macro"
  drm/vmwgfx: Fix LDU X blank screen until mode change issue
  drm/vmwgfx: Skipping fbdev fb pinning for ldu
  drm/vmwgfx: Explicityly track screen target width and height
  drm/vmwgfx: Turn on DRIVER_ATOMIC flag
  drm/vmwgfx: Switch over to internal atomic API for SOU and LDU
  drm/vmwgfx: Switch over to internal atomic API for STDU
  drm/vmwgfx: Fixes to vmwgfx_fb
  drm/vmwgfx: Add and connect atomic state object check/commit
  drm/vmwgfx: Add and connect connector helper function
  drm/vmwgfx: Add and connect plane helper functions
  drm/vmwgfx: Add and connect CRTC helper functions
  drm/vmwgfx: Connector atomic state
  drm/vmwgfx: Plane atomic state
  drm/vmwgfx: CRTC atomic state
  ...

7 years agodrm/vmwgfx: Properly check display/scanout surface size
Sinclair Yeh [Mon, 27 Mar 2017 18:12:27 +0000 (11:12 -0700)]
drm/vmwgfx: Properly check display/scanout surface size

The scanout surface size is the smaller of max texture size and
max STDU size.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Support topology greater than texture size
Sinclair Yeh [Thu, 23 Mar 2017 22:39:16 +0000 (15:39 -0700)]
drm/vmwgfx: Support topology greater than texture size

Most of the display servers today use a single surface to represent the
entire desktop even if it's stretched across multiple screens.

For vmwgfx with STDU, the maximum surface size is limited to the
maximum texture size on the host.  On a 2D VM, this limits our
ability to support configurations with more than one 4K monitor.

To get past this limitation, we will now allow using a large DMA buf
as the framebuffer, and take care of blitting contents from this DMA buf
to the display buffer.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Define an overlaid handle_close ioctl.
Thomas Hellstrom [Thu, 23 Mar 2017 22:26:16 +0000 (15:26 -0700)]
drm/vmwgfx: Define an overlaid handle_close ioctl.

Instead of providing an ioctl for each handle type, provide a single
handle_close ioctl, and reuse the UNREF_DMABUF ioctl.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Sinclair Yeh <syeh@vmware.com>
7 years agodrm/vmwgfx: Re-implement the stream resource as a simple resource.
Thomas Hellstrom [Thu, 23 Mar 2017 22:19:45 +0000 (15:19 -0700)]
drm/vmwgfx: Re-implement the stream resource as a simple resource.

Provide and document a reference implementation.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Sinclair Yeh <syeh@vmware.com>
7 years agodrm/vmwgfx: Introduce a simple resource type
Thomas Hellstrom [Thu, 23 Mar 2017 22:09:42 +0000 (15:09 -0700)]
drm/vmwgfx: Introduce a simple resource type

The callbacks we need to provide to many resources are very similar, so
provide a simple resource type with a number of helpers for these
callbacks.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Sinclair Yeh <syeh@vmware.com>
7 years agodrm/vmwgfx: Revert "drm/vmwgfx: Replace numeric parameter like 0444 with macro"
Øyvind A. Holm [Thu, 23 Mar 2017 21:54:48 +0000 (14:54 -0700)]
drm/vmwgfx: Revert "drm/vmwgfx: Replace numeric parameter like 0444 with macro"

This reverts commit 2d8e60e8b074 ("drm/vmwgfx: Replace numeric
parameter like 0444 with macro")

The commit belongs to the series of 1285 patches sent to LKML on
2016-08-02, it changes the representation of file permissions from the
octal value "0600" to "S_IRUSR | S_IWUSR".

The general consensus was that the changes does not increase
readability, quite the opposite; 0600 is easier to parse mentally than
S_IRUSR | S_IWUSR.

It also causes argument inconsistency, due to commit 04319d89fbec
("drm/vmwgfx: Add an option to change assumed FB bpp") that added
another call to module_param_named() where the permissions are written
as 0600.

Signed-off-by: Øyvind A. Holm <sunny@sunbase.org>
Reviewed-by: Sinclair Yeh <syeh@vmware.com>
7 years agodrm/vmwgfx: Fix LDU X blank screen until mode change issue
Sinclair Yeh [Thu, 23 Mar 2017 21:42:36 +0000 (14:42 -0700)]
drm/vmwgfx: Fix LDU X blank screen until mode change issue

vmw_ldu_crtc_helper_commit() is not called if
drm_atomic_crtc_needs_modeset() decides nothing related to CRTC timing has
changed.

So a better place for this code is in vmw_ldu_primary_plane_atomic_update()
since we will need to update ld->fb every time the FB is updated.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Skipping fbdev fb pinning for ldu
Sinclair Yeh [Thu, 23 Mar 2017 21:41:21 +0000 (14:41 -0700)]
drm/vmwgfx: Skipping fbdev fb pinning for ldu

Pinning fbdev's FB at the start of VRAM prevents X from pinning
its FB.  Since for ldu, the fb would be pinned anyway during a
mode set, just skip pinning it in fbdev.

This is not the best solution, but since ldu is not used much
anymore, it seems like a reasonable workaround.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Explicityly track screen target width and height
Sinclair Yeh [Thu, 23 Mar 2017 21:40:04 +0000 (14:40 -0700)]
drm/vmwgfx: Explicityly track screen target width and height

We can no longer make the assumption that vmw_stdu_update_st() will
be called when there's a valid display surface attached.  So
instead of using display_srf for width and height, make a record of
these paremeters when the screen target is first defined.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Turn on DRIVER_ATOMIC flag
Sinclair Yeh [Fri, 31 Mar 2017 17:16:22 +0000 (10:16 -0700)]
drm/vmwgfx: Turn on DRIVER_ATOMIC flag

Now that the legacy path has been tested, turn on the
DRIVER_ATOMIC flag so user mode driver can start going through
the Atomic path.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
7 years agodrm/vmwgfx: Switch over to internal atomic API for SOU and LDU
Sinclair Yeh [Thu, 23 Mar 2017 21:38:18 +0000 (14:38 -0700)]
drm/vmwgfx: Switch over to internal atomic API for SOU and LDU

Switch over to internal atomic API.  This completes the atomic
internal atomic switch for all the Display Units.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Switch over to internal atomic API for STDU
Sinclair Yeh [Thu, 23 Mar 2017 21:29:22 +0000 (14:29 -0700)]
drm/vmwgfx: Switch over to internal atomic API for STDU

Switch over to using internal atomic API for mode set.

This removes the legacy set_config API, replacing it with
drm_atomic_helper_set_config().  The DRM helper will use various
vmwgfx-specific atomic functions to set a mode.

DRIVER_ATOMIC capability flag is not yet set, so the user mode
will still use the legacy mode set IOCTL.

v2:
* Avoid a clash between page-flip pinning and setcrtc pinning, modify
the page-flip code to use the page-flip helper and the atomic callbacks.
To enable this, we will need to add a wrapper around atomic_commit.

* Add vmw_kms_set_config() to work around vmwgfx xorg driver bug

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Fixes to vmwgfx_fb
Sinclair Yeh [Thu, 23 Mar 2017 21:28:21 +0000 (14:28 -0700)]
drm/vmwgfx: Fixes to vmwgfx_fb

1.  When unsetting a mode, num_connector should be set to zero
2.  The pixel_format field needs to be initialized as newer DRM internal
    functions checks this field
3.  Take the drm_modeset_lock_all() because vmw_fb_kms_detach() can
    change current mode

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Add and connect atomic state object check/commit
Sinclair Yeh [Thu, 23 Mar 2017 21:24:53 +0000 (14:24 -0700)]
drm/vmwgfx: Add and connect atomic state object check/commit

This connects the main state object check and commit function.

v2
* Use drm_atomic_helper_commit() rather than a vmwgfx-specific one

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
7 years agodrm/vmwgfx: Add and connect connector helper function
Sinclair Yeh [Thu, 23 Mar 2017 21:23:20 +0000 (14:23 -0700)]
drm/vmwgfx: Add and connect connector helper function

These helpers won't be called until we flip on the atomic support
flag or set drm_crtc_funcs->set_config to using the atomic
helper.

v2
Use drm_atomic_helper_best_encoder() rather than a vmwgfx-specific one

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
7 years agodrm/vmwgfx: Add and connect plane helper functions
Sinclair Yeh [Thu, 23 Mar 2017 21:18:32 +0000 (14:18 -0700)]
drm/vmwgfx: Add and connect plane helper functions

Refactor previous FB and cursor plane update code into their
atomic counterparts: check, update, prepare, cleanup, and disable.

These helpers won't be called until we flip on the atomic support
flag or set drm_crtc_funcs->set_config to using the atomic
helper.

v2:
* Removed unnecessary pinning of cursor surface
* Added a few function headers

v3:
* Set clip region equal to the destination region
* Fixed surface pinning policy
* Enable SVGA mode in vmw_sou_primary_plane_prepare_fb

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Add and connect CRTC helper functions
Sinclair Yeh [Thu, 23 Mar 2017 20:14:54 +0000 (13:14 -0700)]
drm/vmwgfx: Add and connect CRTC helper functions

Atomic mode set requires us to refactor existing vmw_stdu_crtc_set_config
code into sections that check the validity of the new mode, and sections
that actually program the hardware state.

vmw_du_crtc_atomic_check() takes CRTC-related checking code.  In a later
patch, vmw_du_primary_plane_atomic_check() will take framebuffer-related
checking code.

These helpers won't be called until we flip on the atomic support
flag or set drm_crtc_funcs->set_config to using the atomic
helper.

v2:
* The state->num_connector is actually the total number of potential
  connectors, not just the one associated with the display unit.
  The proper one to check is ->connector_mask.

* Add the check to only allow plane state to be the same as crtc state
  (Thanks to mlankhorst)

* Make sure to turn on SVGA mode before using VRAM.  SVGA mode is
  disabled in master_drop if dbdev is not running.

v3:
* Moved dot clock override to crtc_atomic_check

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
7 years agodrm/vmwgfx: Connector atomic state
Sinclair Yeh [Thu, 23 Mar 2017 18:48:44 +0000 (11:48 -0700)]
drm/vmwgfx: Connector atomic state

Add connector handling functions. Start tracking is_implicity in
the connector state.  Eventually, this field should be tracked
exclusively in a connector state.

Now that plane and connector states have been created, we can also
activate the code that use CRTC state.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Plane atomic state
Sinclair Yeh [Thu, 23 Mar 2017 18:36:05 +0000 (11:36 -0700)]
drm/vmwgfx: Plane atomic state

Add plane state handling functions.

We have to keep track of a few plane states so we cannot use the
DRM helper for this.

Created vmw_plane_state along with functions to reset, duplicate,
and destroty it.

v2
* Removed cursor clean up special case

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
7 years agodrm/vmwgfx: CRTC atomic state
Sinclair Yeh [Thu, 23 Mar 2017 18:33:39 +0000 (11:33 -0700)]
drm/vmwgfx: CRTC atomic state

Create and Add CRTC state.  We currently do not track any properties
or custom states so we can technically use the DRM helpers.  Creating
this code just to make potential future additions easier.

Most of the new code will be compiled but not enabled until
plane/connector state handling code is also in place.

This is the first of a series to enable atomic mode set for vmwgfx.

The atomic enabling effort was done in collaboration with Thomas
Hellstrom and the VMware Graphics Team.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Add universal plane support
Sinclair Yeh [Thu, 23 Mar 2017 18:28:11 +0000 (11:28 -0700)]
drm/vmwgfx: Add universal plane support

Universal support is prerequisite for atomic mode set.

Explicitly create planes for the cursor and the primary FB.  With
a functional cursor plane, the DRM will no longer use the legacy
cursor_set2 and cursor_move entry points.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agodrm/vmwgfx: Removed unused snooper.crtc field
Sinclair Yeh [Thu, 23 Mar 2017 16:59:26 +0000 (09:59 -0700)]
drm/vmwgfx: Removed unused snooper.crtc field

This field is not being used anymore

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
7 years agoMerge branch 'drm-next-4.12' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Fri, 31 Mar 2017 01:47:18 +0000 (11:47 +1000)]
Merge branch 'drm-next-4.12' of git://people.freedesktop.org/~agd5f/linux into drm-next

New stuff for 4.12:
- Preliminary vega10 support
- Support for multi-level page tables
- GPU sensor stuff for mesa
- job tracing improvements
- PRT support for sparse buffers
- Additional SR-IOV improvements
- ttm improvements
- misc bug fixes and code cleanups

* 'drm-next-4.12' of git://people.freedesktop.org/~agd5f/linux: (315 commits)
  drm/amdgpu: Fix 32bit x86 compilation warning
  drm/amdgpu: just disallow reading untouched registers
  drm/amdgpu: remove duplicate allowed reg CP_CPF_BUSY_STAT
  drm/amdgpu/soc15: enable psp block for SRIOV
  drm/amdgpu/soc15: bypass pp block for vf
  drm/amdgpu/psp: add check sOS sign
  drm/amd/amdgpu: Correct ring wptr address in debugfs (v2)
  drm/amdgpu: Fix multi-level page table bugs for large BOs v3
  drm/amdgpu: Fix Vega10 VM initialization
  drm/amdgpu: Make max_pfn 64-bit
  drm/amdgpu: drop GB_GPU_ID from the golden settings
  drm/amdgpu: fix vm pte pde flags to 64-bit for sdma (v3)
  drm/amd/amdgpu: fix Tonga S3 resume hang on rhel6.8
  drm/ttm: decrease ttm bo priority number
  drm/amd/amdgpu: fix performance drop when VRAM pressure
  drm/amdgpu: Couple small warning fixes
  drm/amdgpu: Clean up GFX 9 VM fault messages
  drm/amdgpu: Register UTCL2 as a source of VM faults
  drm/amdgpu/soc15: drop support for reading some registers
  drm/amdgpu/soc15: return cached values for some registers (v2)
  ...

7 years agoapple-gmux: Don't switch external DP port on 2011+ MacBook Pros
Lukas Wunner [Fri, 10 Mar 2017 20:23:45 +0000 (21:23 +0100)]
apple-gmux: Don't switch external DP port on 2011+ MacBook Pros

On MacBook Pros introduced 2011 and onward, external DP ports are
combined DP/Thunderbolt ports that are no longer fully switchable
between GPUs, they can only be driven by the discrete GPU.

More specifically, the Main Link pins (which transport the actual video
and audio streams) are soldered to the discrete GPU, whereas the AUX
Channel pins are switchable. Because the integrated GPU is missing the
Main Link, external displays appear to it as phantoms which fail to
link-train.

Force the AUX channel to the discrete GPU on these models to avoid any
confusion. Document the switching policy implemented by this commit.

Acked-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: http://patchwork.freedesktop.org/patch/msgid/4d1fcc92d1960049e2cff997fbd2d74e45e84e49.1489145162.git.lukas@wunner.de
7 years agodrm/nouveau: Don't register Thunderbolt eGPU with vga_switcheroo
Lukas Wunner [Fri, 10 Mar 2017 20:23:45 +0000 (21:23 +0100)]
drm/nouveau: Don't register Thunderbolt eGPU with vga_switcheroo

An external Thunderbolt GPU can neither drive the laptop's panel nor be
powered off by the platform, so there's no point in registering it with
vga_switcheroo.  In fact, when the external GPU is runtime suspended,
vga_switcheroo will cut power to the internal discrete GPU, resulting in
a lockup.

Cc: Ben Skeggs <bskeggs@redhat.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: http://patchwork.freedesktop.org/patch/msgid/8e733152b13e7c14501ad5af45c1c5c736584111.1489145162.git.lukas@wunner.de
7 years agodrm/amdgpu: Don't register Thunderbolt eGPU with vga_switcheroo
Lukas Wunner [Fri, 10 Mar 2017 20:23:45 +0000 (21:23 +0100)]
drm/amdgpu: Don't register Thunderbolt eGPU with vga_switcheroo

An external Thunderbolt GPU can neither drive the laptop's panel nor be
powered off by the platform, so there's no point in registering it with
vga_switcheroo.  In fact, when the external GPU is runtime suspended,
vga_switcheroo will cut power to the internal discrete GPU, resulting in
a lockup.  Moreover AMD's Windows driver special-cases Thunderbolt as
well.

Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: http://patchwork.freedesktop.org/patch/msgid/701a8e89ce8ac39734736ab779558b6a4042a19e.1489145162.git.lukas@wunner.de
7 years agodrm/radeon: Don't register Thunderbolt eGPU with vga_switcheroo
Lukas Wunner [Fri, 10 Mar 2017 20:23:45 +0000 (21:23 +0100)]
drm/radeon: Don't register Thunderbolt eGPU with vga_switcheroo

An external Thunderbolt GPU can neither drive the laptop's panel nor be
powered off by the platform, so there's no point in registering it with
vga_switcheroo.  In fact, when the external GPU is runtime suspended,
vga_switcheroo will cut power to the internal discrete GPU, resulting in
a lockup.  Moreover AMD's Windows driver special-cases Thunderbolt as
well.

Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: http://patchwork.freedesktop.org/patch/msgid/72d8a9645aece3eff44e116303f0fec8be061c88.1489145162.git.lukas@wunner.de
7 years agoPCI: Recognize Thunderbolt devices
Lukas Wunner [Fri, 10 Mar 2017 20:23:45 +0000 (21:23 +0100)]
PCI: Recognize Thunderbolt devices

Detect on probe whether a PCI device is part of a Thunderbolt controller.
Intel uses a Vendor-Specific Extended Capability (VSEC) with ID 0x1234
on such devices.  Detect presence of this VSEC and cache it in a newly
added is_thunderbolt bit in struct pci_dev.

Also, add a helper to check whether a given PCI device is situated on a
Thunderbolt daisy chain (i.e., below a PCI device with is_thunderbolt
set).

The necessity arises from the following:

* If an external Thunderbolt GPU is connected to a dual GPU laptop,
  that GPU is currently registered with vga_switcheroo even though it
  can neither drive the laptop's panel nor be powered off by the
  platform.  To vga_switcheroo it will appear as if two discrete
  GPUs are present.  As a result, when the external GPU is runtime
  suspended, vga_switcheroo will cut power to the internal discrete GPU
  which may not be runtime suspended at all at this moment.  The
  solution is to not register external GPUs with vga_switcheroo, which
  necessitates a way to recognize if they're on a Thunderbolt daisy
  chain.

* Dual GPU MacBook Pros introduced 2011+ can no longer switch external
  DisplayPort ports between GPUs.  (They're no longer just used for DP
  but have become combined DP/Thunderbolt ports.)  The driver to switch
  the ports, drivers/platform/x86/apple-gmux.c, needs to detect presence
  of a Thunderbolt controller and, if found, keep external ports
  permanently switched to the discrete GPU.

v2: Make kerneldoc for pci_is_thunderbolt_attached() more precise,
    drop portion of commit message pertaining to separate series.
    (Bjorn Helgaas)

Cc: Andreas Noever <andreas.noever@gmail.com>
Cc: Michael Jamet <michael.jamet@intel.com>
Cc: Tomas Winkler <tomas.winkler@intel.com>
Cc: Amir Levy <amir.jer.levy@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: http://patchwork.freedesktop.org/patch/msgid/0ab165a4a35c0b60f29d4c306c653ead14fcd8f9.1489145162.git.lukas@wunner.de
7 years agoMAINTAINERS: Add Lukas Wunner as reviewer for vga_switcheroo
Lukas Wunner [Mon, 27 Mar 2017 16:48:24 +0000 (18:48 +0200)]
MAINTAINERS: Add Lukas Wunner as reviewer for vga_switcheroo

I've been contributing to vga_switcheroo for the past two years and by
now am fairly familiar with it, so danvet suggested that I add myself
as reviewer.

While at it, add missing file pattern for vga_switcheroo.h + vgaarb.h
to the DRM and DRM-MISC sections such that get_maintainer.pl returns
dri-devel@ and the drm-misc maintainers.

Suggested-and-acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: http://patchwork.freedesktop.org/patch/msgid/ff2320a0790d039e714cf352cf32ec16fa370627.1490623913.git.lukas@wunner.de
7 years agodrm: Fix locking gotcha in page_flip ioctl
Daniel Vetter [Thu, 30 Mar 2017 20:48:31 +0000 (22:48 +0200)]
drm: Fix locking gotcha in page_flip ioctl

We want to lock the primary plane, not the cursor (which might be
optional). Real bad case of copy-paste fail, unfortunately our CI
didn't catch that because i915 does have a cursor plane.

Reported-by: Eric Anholt <eric@anholt.net>
Fixes: 29dc0d1de182 ("drm: Roll out acquire context for the page_flip ioctl")
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170330204831.8225-1-daniel.vetter@ffwll.ch
Tested-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
7 years agodrm/amdgpu: Fix 32bit x86 compilation warning
Alex Xie [Thu, 30 Mar 2017 17:30:00 +0000 (13:30 -0400)]
drm/amdgpu: Fix 32bit x86 compilation warning

drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c:187:2: warning: right shift count >= width of type [enabled by default]
drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c:173:2: warning: right shift count >= width of type [enabled by default]
drivers/gpu/drm/amd/amdgpu/vega10_ih.c:106:3: warning: right shift count >= width of type [enabled by default]

v2: Add a space between "&" and "0xff"

Reported by: kbuild-all@01.org

Signed-off-by: Alex Xie <AlexBin.Xie@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm: Clarify the role of plane_state argument to drm_simple update().
Eric Anholt [Mon, 20 Mar 2017 23:36:15 +0000 (16:36 -0700)]
drm: Clarify the role of plane_state argument to drm_simple update().

Like the atomic update hook it's wrapping, the plane_state is the old
one, and the new one is in plane->state.  Both msxfb and tinydrm use
it correctly, but I mistook it for the new state in pl111 due to its
naming.

Signed-off-by: Eric Anholt <eric@anholt.net>
Link: http://patchwork.freedesktop.org/patch/msgid/20170320233615.5242-3-eric@anholt.net
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
7 years agodrm/amdgpu: just disallow reading untouched registers
Christian König [Fri, 24 Mar 2017 13:32:50 +0000 (14:32 +0100)]
drm/amdgpu: just disallow reading untouched registers

Not sure what the original intention was here, but returning a random piece of
kernel memory to userspace because we didn't set the value at all is clearly
not a good idea.

This patch disallows reading the register and returns
a proper error code instead.

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Tom St Denis <tom.stdenis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: remove duplicate allowed reg CP_CPF_BUSY_STAT
Christian König [Fri, 24 Mar 2017 13:05:37 +0000 (14:05 +0100)]
drm/amdgpu: remove duplicate allowed reg CP_CPF_BUSY_STAT

Remove duplicate mmCP_CPF_BUSY_STAT from the allowed registers.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm: Clear e after kfree in drm_mode_page_flip_ioctl
Daniel Vetter [Thu, 30 Mar 2017 13:32:53 +0000 (15:32 +0200)]
drm: Clear e after kfree in drm_mode_page_flip_ioctl

With the explicit retry loop static analyzers get confused by the
control flow and believe that e could be accessed after kfree. That's
not possible, but it's non-obvious, so let's clear it to NULL.

We already cleared e = NULL at the top of the function, so this is all
in line.

Cc: Julia Lawall <julia.lawall@lip6.fr>
Reported-by: Julia Lawall <julia.lawall@lip6.fr>
Fixes: 29dc0d1de182 ("drm: Roll out acquire context for the page_flip ioctl")
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170330133253.29500-1-daniel.vetter@ffwll.ch
7 years agodrm: Convert cmpxchg(bool) back to a two step operation
Chris Wilson [Thu, 30 Mar 2017 14:08:32 +0000 (15:08 +0100)]
drm: Convert cmpxchg(bool) back to a two step operation

ARM v6 (at least) only allows cmpxchg on 32bit variables which doesn't
always include the bool type.

drivers/built-in.o: In function `vblank_disable_and_save':
imx-ocotp.c:(.text+0xb45e8): undefined reference to `__bad_cmpxchg'
Makefile:986: recipe for target 'vmlinux' failed

Reported-by: kbuild test robot <fengguang.wu@intel.com>
Reported-by: Leonard Crestez <leonard.crestez@nxp.com>
Fixes: 43dc7fe2b211 ("drm: Mark up accesses of vblank->enabled outside of its spinlock")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170330140832.32377-1-chris@chris-wilson.co.uk
7 years agodrm/bridge: ti-tfp410: support hpd via gpio
Christopher Spinrath [Mon, 6 Mar 2017 21:40:43 +0000 (22:40 +0100)]
drm/bridge: ti-tfp410: support hpd via gpio

On some boards the hpd pin of a hdmi connector is wired up to a gpio
pin. Since in the DRM world the tfp410 driver is responsible for
handling the connector, add support for hpd gpios in this very driver.

Reviewed-by: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Christopher Spinrath <christopher.spinrath@rwth-aachen.de>
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Link: http://patchwork.freedesktop.org/patch/msgid/2e47786ab3d04078ae70d0c4064f7c4d@rwthex-s1-b.rwth-ad.de
7 years agodrm: use .hword to represent 16-bit numbers
Javi Merino [Wed, 29 Mar 2017 13:45:23 +0000 (14:45 +0100)]
drm: use .hword to represent 16-bit numbers

The size of .word is the size of a word in the given platform, which
for intel systems is 16-bits but other architectures use different
sizes.  However, .hword emits 16-bit numbers regardless of the
platform (and despite the name).  The quantities specified in EDID are
platform independent, so they should work in spite of the default
target of the cc you are using, so use .hword where EDID specifies
16-bit numbers.

Cc: Carsten Emde <C.Emde@osadl.org>
Cc: David Airlie <airlied@linux.ie>
Acked-by: David Airlie <airlied@linux.ie>
Signed-off-by: Javi Merino <javi.merino@kernel.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1490795123-16851-1-git-send-email-javi.merino@kernel.org
7 years agoRevert unrelated part of "drm: simplify the locking in the GETCRTC ioctl"
Maarten Lankhorst [Thu, 30 Mar 2017 07:36:05 +0000 (09:36 +0200)]
Revert unrelated part of "drm: simplify the locking in the GETCRTC ioctl"

v2 of the commit 2c77bb29d398 ("drm: simplify the locking in the GETCRTC ioctl")
accidentally introduced a unrelated change in intel_display.c, revert the
unrelated change.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Fixes: 2c77bb29d398 ("drm: simplify the locking in the GETCRTC ioctl")
Reported-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/6be47261-475f-c190-af56-c136677246d9@linux.intel.com
7 years agodrm: Fixup failure paths in drm_atomic_helper_set_config
Daniel Vetter [Wed, 29 Mar 2017 17:41:36 +0000 (19:41 +0200)]
drm: Fixup failure paths in drm_atomic_helper_set_config

I've screwed this up when removing the legacy backoff hack.

Fixes: 38b6441e4e75 ("drm/atomic-helper: Remove the backoff hack from set_config")
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Cc: David Airlie <airlied@linux.ie>
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170329174136.10330-1-daniel.vetter@ffwll.ch
7 years agodrm/amdgpu/soc15: enable psp block for SRIOV
Xiangliang Yu [Tue, 28 Mar 2017 11:16:42 +0000 (19:16 +0800)]
drm/amdgpu/soc15: enable psp block for SRIOV

SRIOV can support for loading ucode with PSP block, enable it.

Signed-off-by: Xiangliang Yu <Xiangliang.Yu@amd.com>
Acked-by: Huang Rui <ray.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu/soc15: bypass pp block for vf
Xiangliang Yu [Tue, 28 Feb 2017 09:26:40 +0000 (17:26 +0800)]
drm/amdgpu/soc15: bypass pp block for vf

Disable pp block if device is  vf.

Signed-off-by: Xiangliang Yu <Xiangliang.Yu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Monk Liu <Monk.Liu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu/psp: add check sOS sign
Xiangliang Yu [Tue, 28 Mar 2017 11:48:10 +0000 (19:48 +0800)]
drm/amdgpu/psp: add check sOS sign

Confirm if sys driver and sOS are already been loaded through sOS
sign register, skip loading sys driver and sOS if finding the sign.

Signed-off-by: Xiangliang Yu <Xiangliang.Yu@amd.com>
Acked-by: Huang Rui <ray.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amd/amdgpu: Correct ring wptr address in debugfs (v2)
Tom St Denis [Wed, 29 Mar 2017 17:01:30 +0000 (13:01 -0400)]
drm/amd/amdgpu: Correct ring wptr address in debugfs (v2)

On gfx9 hardware the value is not wrapped and is a 64-bit value.  So
we reduce it modulo the ring size.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
(v2) use buf_mask instead of computing on the fly

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Fix multi-level page table bugs for large BOs v3
Felix Kuehling [Wed, 29 Mar 2017 00:36:12 +0000 (20:36 -0400)]
drm/amdgpu: Fix multi-level page table bugs for large BOs v3

Fix the start/end address calculation for address ranges that span
multiple page directories in amdgpu_vm_alloc_levels.

Add error messages if page tables aren't found. Otherwise the page
table update would just fail silently.

v2:
 * Change WARN_ON to WARN_ON_ONCE
 * Move masking of high address bits to caller
 * Add range-check for "from" and "to"
v3:
 * Replace WARN_ON_ONCE in get_pt with pr_err in caller

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Fix Vega10 VM initialization
Felix Kuehling [Wed, 29 Mar 2017 00:31:20 +0000 (20:31 -0400)]
drm/amdgpu: Fix Vega10 VM initialization

adev->family is not initialized yet when amdgpu_get_block_size is
called. Use adev->asic_type instead.

Minimum VM size is 512GB, not 256GB, for a single page table entry
in the root page table.

gmc_v9_0_vm_init is called after adev->vm_manager.max_pfn is
initialized. Move the minimum VM-size enforcement ahead of max_pfn
initializtion. Cast to 64-bit before the left-shift.

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: Make max_pfn 64-bit
Felix Kuehling [Wed, 29 Mar 2017 00:24:53 +0000 (20:24 -0400)]
drm/amdgpu: Make max_pfn 64-bit

With 4-level page tables the maximum VM size is 256TB. That's 64G
pages, which can't be represented in 32-bit.

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agodrm/amdgpu: drop GB_GPU_ID from the golden settings
Christian König [Wed, 22 Mar 2017 09:29:45 +0000 (10:29 +0100)]
drm/amdgpu: drop GB_GPU_ID from the golden settings

That register is marked deprecated, reading it results in a bus error.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>