move virtual machine related structure to amdgpu_virt.h
easy for developer to maintain for virualization stuffs
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The check for an out of bound index into array interrupt_status_offsets
is off-by-one. Fix this and also don't compared to a hard coded array
size but use adev->mode_info.num_hpd instead.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We get 7 warnings when building kernel with W=1:
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c:1990:5: warning: no previous prototype for 'amdgpu_pre_soft_reset' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c:1548:5: warning: no previous prototype for 'amdgpu_connector_virtual_dpms' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c:1560:5: warning: no previous prototype for 'amdgpu_connector_virtual_set_property' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c:330:5: warning: no previous prototype for 'amdgpu_cs_list_validate' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/dce_virtual.c:98:6: warning: no previous prototype for 'dce_virtual_stop_mc_access' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/dce_virtual.c:130:6: warning: no previous prototype for 'dce_virtual_resume_mc_access' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/dce_virtual.c:136:6: warning: no previous prototype for 'dce_virtual_set_vga_render_state' [-Wmissing-prototypes]
In fact, all of the functions are only used in the file
in which they are declared and don't need a declaration,
but can be made static.
So this patch marks both functions with 'static'.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Baoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We get 2 warnings when building kernel with W=1:
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c:146:5: warning: no previous prototype for 'pool_to_domain' [-Wmissing-prototypes]
drivers/gpu/drm/amd/amdgpu/cz_smc.c:104:5: warning: no previous prototype for 'cz_send_msg_to_smc_with_parameter_async' [-Wmissing-prototypes]
In fact, both functions are called by no one and not exported,
so this patch removes them.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Baoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Replace per-asic print_current_performance() functions with generic
that calls read_sensor. Tested on Tonga and Carrizo for aesthetics
and accuracy.
Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit aff51175cd.
The commit caused fence timeouts within nvc0_screen_destroy and most likely
other places as well.
The most obvious effect is, that userspace processes take minutes to
actually quit.
Signed-off-by: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The highest 3bits of the 4 layers buffers are all part of the same
register. However, our mask computation was wrong, leading to all the
lowest register bits being removed when we use regmap_update_bits, which
will lead to the buffers being set to some random part of the RAM.
Fix our mask.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
The drm_mode_create call might return NULL in case of a failure, and the
current code doesn't check for that. Make sure it does.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Even though all our planes can support the ARGB formats, the lowest plane
(ie the primary plane) cannot use the alpha component, otherwise it will
just result in the transparent area being entirely black.
Since some applications will still require the ARGB format, let's force the
format to XRGB to drop the alpha component entirely.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Just pure code movement, cleanup and polish will happen in later
patches.
v2: Don't forget all the ioctl! To extract those cleanly I decided to
put check_src_coords into drm_framebuffer.c (and give it a
drm_framebuffer_ prefix), since that just checks framebuffer
constraints.
v3: rebase over PAGE_FLIP_TARGET.
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
[seanpaul]
This patch as posted on the list was rebased on:
commit 6f00975c61
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Sat Aug 20 12:22:11 2016 +0200
drm: Reject page_flip for !DRIVER_MODESET
so as a result of moving the page_flip ioctl, this fix has
been rolled into this patch.
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Also reorder alphabetically and fix up drm_flip_work header.
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
With DP MST, a digital_port can carry more than one audio stream. Hence,
more than one audio_connector needs to be attached to intel_digital_port in
such cases. However, each stream is associated with an unique encoder. So,
instead of creating an array of audio_connectors per port, move
audio_connector from struct intel_digital_port to struct intel_encoder.
This also simplifies access to the right audio_connector from codec
functions in intel_audio.c that receive intel_encoder.
v2: Removed locals that are not needed anymore.
v3: No code change except for minor change in context.
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Reviewed-by: Lyude <cpaul@redhat.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1474334681-22690-5-git-send-email-dhinakaran.pandiyan@intel.com
There is a disparity in the context image saved to disk and our own
bookkeeping - that is we presume the RING_HEAD and RING_TAIL match our
stored ce->ring->tail value. However, as we emit WA_TAIL_DWORDS into the
ring but may not tell the GPU about them, the GPU may be lagging behind
our bookkeeping. Upon hibernation we do not save stolen pages, presuming
that their contents are volatile. This means that although we start
writing into the ring at tail, the GPU starts executing from its HEAD
and there may be some garbage in between and so the GPU promptly hangs
upon resume.
Testcase: igt/gem_exec_suspend/basic-S4
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=96526
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20160921135108.29574-3-chris@chris-wilson.co.uk
If create_crtc fails in vop bind, ensure the irq refcount is zeroed
back out before exiting.
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
After changes introduced by last patches, there is no useful data stored
in vop_plane_state struct. Let's remove it and make the driver use
generic plane state alone.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
This patch makes the driver send the pending vblank event in next vblank
following the commit, relying on vblank signalling improvements done in
previous patches. This gives us vblank events that always represent the
real moment of changes hitting on the screen (which was the case only
for complete FB changes before) and lets us remove the manual window
update check.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Originally we needed to enable vblank for any atomic commit to kick the
PSR machine, but that was changed and we no longer need to do so from
a vblank interrupt. Let's return to original behavior of enabling
vblank only if it is really necessary.
This essentially reverts commit 5b6804034a ("drm/rockchip: Enable
vblank without event").
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Currently the driver uses a custom function to wait for flip to complete
after an atomic commit. It was needed before because of two problems:
- there is no hardware vblank counter, so the original helper would
have a race condition with the vblank interrupt,
- the driver didn't support unreferencing cursor framebuffers
asynchronously to the commit, which was what the helper expected.
Since both problems have been solved by previous patches, we can now
make the driver use the generic helper and remove custom waiting code.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Currently the driver waits for vblank and then unreferences old
framebuffers from atomic commit code path. This is however breaking the
legacy cursor API, which requires the updates to be fully asynchronous.
Instead of just adding a special case for cursor, we can have actually
smaller amount of code to unreference any changed framebuffer from a
flip work.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Since VOP does not have a hardware vblank count register, the ongoing
commit might be racing with a requested vblank interrupt, which would
increment the software vblank counter before the changes being committed
actually happen.
To avoid this, we can extend .atomic_flush(), so after it sets cfg_done
bit, it polls the vblank interrupt bit until it's inactive to make sure
that any old vblank interrupt gets to the handler and then uses
synchronize_irq(vop->irq) to make sure the handler finishes running.
The polling case should happen very rarely, but even if, the total wait
time should be relatively low and in practice almost equal to the vop
hardirq handler running time.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Current code implements prepare_fb and cleanup_fb callbacks only to
grab/release fb references, which is already done by atomic framework
when creating/destryoing plane state. Let's remove these
unused bits.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
The enable register only masks the raw status bits to signal CPU
interrupt only for enabled interrupts. The status bits are activated
regardless of the enable register. This means that we might have an old
interrupt event queued, which we are not interested in. To avoid getting
a spurious interrupt signalled, we have to clear the old bit before we
update the enable register.
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
The ->set() callback would always be called when transitioning
from FLUSH->DISABLE since we assign state to psr->state right
above the skip condition.
Reported-by: Daniel Kurtz <djkurtz@chromium.org>
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
3 seconds is a bit too conservative, drop this to 100ms for
better power savings.
Reviewed-by: Yakir Yang <ykk@rock-chips.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Instead of keying off vblank for psr, just flush every time
we get an atomic update. This ensures that cursor updates
will properly disable psr (without turning vblank on/off),
and unifies the paths between fb_dirty and atomic psr
enable/disable.
Reviewed-by: Yakir Yang <ykk@rock-chips.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
