When we return pages to the system, we release control over them and
should defensively return them to the CPU write domain so that we catch
any external writes on reacquiring them (e.g. to transparently
swapout/swapin). While we did this defensive clflushing for ordinary
shmem pages, it was forgotten for userptr. Fortunately, userptr objects
are normally cache coherent and so oblivious to the forgotten domain
tracking.
References: a679f58d05 ("drm/i915: Flush pages on acquisition")
References: 754a254427 ("drm/i915: Skip object locking around a no-op set-domain ioctl")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190331094620.15185-1-chris@chris-wilson.co.uk
As we only set ctx->file_priv on registering the GEM context after
construction, it is invalid to try and use it in the middle for setting
various parameters. Indeed, we put the file_priv into struct create_ext
so that we have the right file_private available without having to look
at ctx->file_priv. However, it helps to use it!
Reported-by: Jordan Justen <jordan.l.justen@intel.com>
Fixes: b917154172 ("drm/i915: Extend CONTEXT_CREATE to set parameters upon construction")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Tested-by: Jordan Justen <jordan.l.justen@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190330100349.30642-1-chris@chris-wilson.co.uk
Splits out the dither register bits and introduces
the same config enumerations as in the rockchip kernel tree.
Tested to fix the banding on my ASUS C201.
Signed-off-by: Urja Rannikko <urjaman@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20190318154412.26994-1-urjaman@gmail.com
Currently we only retry to load GuC firmware if the load fails due to
timeout. On Gen9 GuC loading may fail for different reasons, not just
hang/timeout. Direction from the GuC team is to retry for all cases of
GuC load failure on Gen9, not just for timeout.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108593
Signed-off-by: Robert M. Fosha <robert.m.fosha@intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190329231746.9129-1-robert.m.fosha@intel.com
EXT2 GC MAX registers are introduced from Gen10+ to
program values from 3.0 to 7.0. Enabled the same, but
currently limiting it to 1.0 as userspace ABI is limited
at that currently.
v2: Updated the 1.0 programming and aligned as per GLK, also added
GLK along with GEN10+ check, as per Ville's feedback.
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1553869756-4546-3-git-send-email-uma.shankar@intel.com
GC MAX register is used to program values from 1.0 to
less than 3.0. A different register was used instead of
the intended one. Fixed the same.
Currently limiting it to 1.0 due to ABI limitations.
v2: Updated the 1.0 programming and aligned as per GLK, based
on Ville's feedback.
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1553869756-4546-2-git-send-email-uma.shankar@intel.com
Those functions are used on gen4 as well and gen4 does have a non-RCS
engine, so remove the BUG_ON and flip back the logic to what it was
before the ENGINE_READ/WRITE update
v2: update the posting read as well (Chris, Ville).
Fixes: baba6e572b ("drm/i915: take a reference to uncore in the engine and use it")
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190329165018.32953-1-daniele.ceraolospurio@intel.com
edram is not part of uncore and there is no requirement for the
detection to be done before we initialize the uncore functions. The
first check on HAS_EDRAM is in the ggtt_init path, so move it to
i915_driver_init_hw, where other dram-related detection happens.
While at it, save the size in MB instead of the capabilities because the
size is the only thing we look at outside of the init function.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190328174533.31532-1-daniele.ceraolospurio@intel.com
Allow atomic_enable and atomic_disable operations from
drm_crtc_helper_funcs struct optional. With this, the target display
drivers don't need to define a dummy function if they don't need one.
Changes since v2:
* Don't make funcs optional
* Update kerneldoc for atomic_enable/disable
* Replace "if (funcs->atomic_enable)" by "if (funcs->commit)"
* Improve commit message
Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20190314184845.gjmvkamobj4dilyp@smtp.gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
According to Intel GFX PRM on 01.org, plane surface address can be updated
synchronously or asynchronously. Synchronous flip will hold plane surface
address update to start of next vsync, which is current implementation.
Asynchronous flip will update the address as soon as possible. Without
async flip, some 3D application could not reach better performance and
the maximum performance is no higher than vsync frequency.
The patch enables the async flip on plane surface address mmio update,
and increment flip count correctly.
With async flip enabled, some 3D applications have significant performance
improvement. i.e. 3DMark Ice Storm has a 300%~400% increment on score.
v2:
Use bit operation definition for flip mode. (zhenyu)
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Colin Xu <colin.xu@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
According to Intel GFX PRM on 01.org, the MI_DISPLAY_FLIP command can
either request display plane flip synchronously or asynchronously.
In synchronous flip, flip will be hold until next vsync, which
is not implemented yet in GVT. In asynchronous flip, flip will happen
immediately, which is current implementation.
The patch enables the sync flip on handling MI_DISPLAY_FLIP,
and increment flip count correctly by only increment on primary plane.
v2:
Use bit operation definition for flip mode. (zhenyu)
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Colin Xu <colin.xu@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Add SKL_FLIP_EVENT to address into intel_gvt_event_type for primary
and sprite0 plane flip event.
Add macro to address REG_50080 offset.
v2:
Add bit operation definition for flip mode. (zhenyu)
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Colin Xu <colin.xu@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
in init_skil_mmio_info, replaced register address with the known
name from i915_reg.h definition to improve code readbility.
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Xiaolin Zhang <xiaolin.zhang@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
shadow mm's pin count got increased in workload preparation phase, which
is after workload scanning.
it will get decreased in complete_current_workload() anyway after
workload completion.
Sometimes, if a workload meets a scanning error, its shadow mm pin count
will not get increased but will get decreased in the end.
This patch lets shadow mm's pin count not go below 0.
Fixes: 2707e44466 ("drm/i915/gvt: vGPU graphics memory virtualization")
Cc: zhenyuw@linux.intel.com
Cc: stable@vger.kernel.org #4.14+
Signed-off-by: Yan Zhao <yan.y.zhao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
in workload creation routine, if any failure occurs, do not queue this
workload for delivery. if this failure is fatal, enter into failsafe
mode.
Fixes: 6d76303553 ("drm/i915/gvt: Move common vGPU workload creation into scheduler.c")
Cc: stable@vger.kernel.org #4.19+
Cc: zhenyuw@linux.intel.com
Signed-off-by: Yan Zhao <yan.y.zhao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
- fix mmap range checks
- fix gvt ppgtt mm LRU list access races
- fix selftest error pointer check
- fix a macro definition (pre-emptive for potential further backports)
- fix one AML SKU ULX status
-----BEGIN PGP SIGNATURE-----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=KtRe
-----END PGP SIGNATURE-----
Merge tag 'drm-intel-fixes-2019-03-28' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes
drm/i915 fixes for v5.2-rc3:
- fix mmap range checks
- fix gvt ppgtt mm LRU list access races
- fix selftest error pointer check
- fix a macro definition (pre-emptive for potential further backports)
- fix one AML SKU ULX status
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87sgv6ao7a.fsf@intel.com
Don't load the linear degamma LUT on ICL. The hardware no longer
has any silly linkages between the CSC enable and degamma LUT
enable so the degamma LUT is only needed when it's actually
enabled.
Also add comments to explain the situation on GLK.
v2: Drop useless parens around 1<<16
v3: Add missing const
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190327155045.28446-11-ville.syrjala@linux.intel.com
With everything else moved out of the way only ilk+
remains using _intel_color_check(). Streamline the logic
into ilk_color_check().
v2: Add some comments explaining we that we don't expose
the full hardware capabilities currently (Matt)
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190327155045.28446-9-ville.syrjala@linux.intel.com
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Unlike the earlier platforms GLK has dedicated degamma and gamma
LUTs. And quite curiously the degamma LUT is actually controlled
via the PLANE_COLOR_CTL CSC enable bit. Hence we must compute
gamma_enable and csc_enable differently to pre-GLK platforms.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190327155045.28446-7-ville.syrjala@linux.intel.com
ICL is rather easy when it comes to .color_check() as it
finally provides us with a full color pipeline with
individual knobs for each stage.
We'll also start bypassing each LUT individually when
it is not needed.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190327155045.28446-6-ville.syrjala@linux.intel.com
Since CHV has the CGM unit we require a custom implementation
of .color_check().
This fixes the computation of gamma_enable as previously we
left it enabled even when were using the CGM gamma instead.
Now we turn off the legacy LUT unless it's actually required.
v2: Add some comment explaining the color pipeline (Matt)
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190327155045.28446-5-ville.syrjala@linux.intel.com
The current intel_color_check() is a mess, and worse yet it is
in fact incorrect for several platforms. The hardware has
evolved quite a bit over the years, so let's just go for a clean
split between the platforms by turning this into a vfunc.
The actual work to split it up will follow.
v2: Assign the vfuncs in the order they appear in the
struct (Matt)
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190327155045.28446-3-ville.syrjala@linux.intel.com
GT VEBOX DISABLE is only 4 bits wide but it was using a 8 bits wide
mask, the remaning reserved bits is set to 0 causing 4 more
nonexistent VEBOX engines being detected as enabled, triggering the
BUG_ON() because of mismatch between vebox_mask and newly added
VEBOX_MASK().
[ 64.081621] [drm:intel_device_info_init_mmio [i915]] vdbox enable: 0005, instances: 0005
[ 64.081763] [drm:intel_device_info_init_mmio [i915]] vebox enable: 00f1, instances: 0001
[ 64.081825] intel_device_info_init_mmio:925 GEM_BUG_ON(vebox_mask != ({ unsigned int first__ = (VECS0); unsigned int count__ = (2); ((&(dev_priv)->__info)->engine_mask & (((~0UL) - (1UL << (first__)) + 1) & (~0UL >> (64 - 1 - (first__ + count__ - 1))))) >> first__; }))
[ 64.082047] ------------[ cut here ]------------
[ 64.082054] kernel BUG at drivers/gpu/drm/i915/intel_device_info.c:925!
BSpec: 20680
Fixes: 26376a7e74 ("drm/i915/icl: Check for fused-off VDBOX and VEBOX instances")
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190326230223.26336-1-jose.souza@intel.com
(cherry picked from commit 547fcf9b1c)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
This patch moves the virtio_gpu_cmd_create_resource() call (which
notifies the host about the new resource created) into the
virtio_gpu_object_create() function. That way we can call
virtio_gpu_cmd_create_resource() before ttm_bo_init(), so the host
already knows about the object when ttm initializes the object and calls
our driver callbacks.
Specifically the object is already created when the
virtio_gpu_ttm_tt_bind() callback invokes virtio_gpu_object_attach(),
so the extra virtio_gpu_object_attach() calls done after
virtio_gpu_object_create() are not needed any more.
The fence support for the create ioctl becomes a bit more tricky though.
The code moved into virtio_gpu_object_create() too. We first submit the
(fenced) virtio_gpu_cmd_create_resource() command, then initialize the
ttm object, and finally attach just created object to the fence for the
command in case it didn't finish yet.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-6-kraxel@redhat.com
Add format, width and height fields to the virtio_gpu_object_params
struct. With that in place we can use the parameter struct for
virtio_gpu_cmd_create_resource() calls too.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-4-kraxel@redhat.com
Create virtio_gpu_object_params, use that to pass object parameters to
virtio_gpu_object_create. This is just the first step, followup patches
will add more parameters to the struct. The plan is to use the struct
for all object parameters.
Drop unused "kernel" parameter for virtio_gpu_alloc_object(), it is
unused and always false.
Also drop "pinned" parameter. virtio-gpu doesn't shuffle around
objects, so effecively they all are pinned anyway. Hardcode
TTM_PL_FLAG_NO_EVICT so ttm knows. Doesn't change much for the moment
as virtio-gpu supports TTM_PL_FLAG_TT only so there is no opportunity to
move around objects. That'll probably change in the future though.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-3-kraxel@redhat.com
Drop the dummy ttm backend implementation, add a real one for
TTM_PL_FLAG_TT objects. The bin/unbind callbacks will call
virtio_gpu_object_{attach,detach}, to update the object state
on the host side, instead of invoking those calls from the
move_notify() callback.
With that in place the move and move_notify callbacks are not
needed any more, so drop them.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-2-kraxel@redhat.com
This file prints which features the virtio-gpu device has.
Also add "virtio-gpu-" prefix to the existing fence file,
to make clear this is a driver-specific debugfs file.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190320083612.32067-1-kraxel@redhat.com
Fixes gcc '-Wunused-but-set-variable' warning:
drivers/gpu/drm/virtio/virtgpu_ttm.c: In function 'virtio_gpu_init_mem_type':
drivers/gpu/drm/virtio/virtgpu_ttm.c:117:28: warning:
variable 'vgdev' set but not used [-Wunused-but-set-variable]
drivers/gpu/drm/virtio/virtgpu_ttm.c: In function 'virtio_gpu_bo_swap_notify':
drivers/gpu/drm/virtio/virtgpu_ttm.c:300:28: warning:
variable 'vgdev' set but not used [-Wunused-but-set-variable]
It is never used since introduction in dc5698e80c ("Add virtio gpu driver.")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190325092631.152060-1-yuehaibing@huawei.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
On the D3 SoC the LVDS PHY must be enabled in the same register write
that enables the LVDS output. Skip writing the LVEN bit independently
on that platform, it will be set by the write that sets LVRES.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
The D3 and E3 SoCs have different pixel clock frequency limits for the
LVDS encoder than the other SoCs in the Gen3 family. Adjust the mode
fixup implementation accordingly.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
