IGT currently does not properly commit changes on planes with multiple
possible CRTC's. Set one valid CRTC for each plane for now, plus one
underlay plane on Carizzo and Stoney that is valid for all CRTCs.
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
1) Always call flip ISR, for any surface comit, not only for page flip.
2) For events not signaled in flip ISR, handle them in end of commit_tail.
3)Fix race condition between subsequent atomic calls by removing
current state access AFTER drm_atomic_helper_commit_hw_done
was called. After this call a pending next commit will
swap in his own state into objects and proceed.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This code is remanant of pre atomic age when flip was a
standalone IOCTL.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Today we use special interface for flip because of fear of cuncurency issues
over dc->current_ctx. This should be no longer an issue when flipping on
multiple CRTCs concurently since for fast update (as flip is) no new context
is created and the exsisitng is not destroyed. For full updates case when
removing or adding streams on once CRTC while flipping on another
Adding all current active CRTC's states to the atomic commit in
amdgpu_dm_atomic_check will garntee that any such full update commit
will wait for completion of any outstanding flip.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
There is a hard hang observed during resume from S3 when
the system receives a DP short pulse interrupt. This is
because there are two code paths contending for GPIO
access for AUX channel transactions. One such path is
through amdgpu_dm_display_resume() function which is
invoked from the regular system resume code path. The
other path is through handle_hpd_rx_irq(), which is
invoked in response to system receiving DP short pulse
interrupt. handle_hpd_rx_irq() guards against conflicting
GPIO access using hpd_lock, but the GPIO access from
amdgpu_dm_display_resume() remains unguarded.
This patch makes sure we use hpd_lock inside
amdgpu_dm_display_resume() to avoid race conditions
for GPIO access.
Signed-off-by: Arindam Nath <arindam.nath@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
They could differ between ASIC generations
Signed-off-by: Sylvia Tsai <sylvia.tsai@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Since DC now uses CRTC_VERTICAL_INTERRUPT0 as VBLANK irq trigger
and vblank interrupts actually happen earliest at start of vblank,
instead of a bit before vblank, we no longer need some of the
fudging logic to deal with too early vblank irq handling (grep for
lb_vblank_lead_lines). This itself fixes a pageflip scheduling
bug in DC, caused by uninitialized use of lb_vblank_lead_lines,
with a wrong startup value of 0. Thanks to the new vblank irq
trigger this value of zero is now actually correct for DC :).
A new problem is that vblank irq's race against pflip irq's,
and as both can fire at first line of vblank, it is no longer
guaranteed that vblank irq handling (therefore -> drm_handle_vblank()
-> drm_update_vblank_count()) executes before pflip irq handling
for a given vblank interval when a pageflip completes. Therefore
the vblank count and timestamps emitted to user-space as part of
the pageflip completion event will be often stale and cause new
timestamping and swap scheduling errors in user-space.
This was observed with large frequency on R9 380 Tonga Pro.
Fix this by enforcing a vblank count+timestamp update right
before emitting the pageflip completion event from the pflip
irq handler. The logic in core drm_update_vblank_count() makes
sure that no redundant or conflicting updates happen, iow. the
call turns into a no-op if it wasn't needed for that vblank,
burning a few microseconds of cpu time though.
Successfully tested on AMD R9 380 "Tonga Pro" (VI/DCE 10)
with DC enabled on the current DC staging branch. Independent
measurement of pageflip completion timing with special hardware
measurement equipment now confirms correct pageflip timestamps
and counts in the pageflip completion events.
v2: Review comments by Michel, drop outdated paragraph
about problem already fixed in 2nd patch of the series.
Add acked/r-b by Harry and Michel.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Acked-by: Harry Wentland <harry.wentland@amd.com> (v1)
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Cc: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
4.9 kernel will always add the assigned crtc to possible_crtcs on a
plane. This is no longer the case on newer kernels. Make sure we allow
any plane on any crtc.
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jordan Lazare <Jordan.Lazare@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Avoid enabling CRTC_VERTICAL_INTERRUPT0 twice on resume.
It's enabled once from within manage_dm_interrupts in mode set
and another explicitly from amdgpu_dm_irq_resume_late.
Seems it lead to CRTC hang.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Current design has per-crtc-plane model.
As a result, for asic's that support underlay,
are unable to expose it to user space for modesetting.
To enable this, the drm driver intialisation now runs
for number of surfaces instead of stream/crtc.
This patch plumbs surface capabilities to drm framework
so that it can be effectively used by user space.
Tests: (On Chromium OS for Stoney Only)
* 'modetest -p' now shows additional plane
with YUV capabilities in case of CZ and ST.
* 'plane_test' fails with below error:
[drm:amdgpu_dm_connector_atomic_set_property [amdgpu]] *ERROR* Unsupported screen depth 0
as ther is no support for YUYV
* Checked multimonitor display works fine
Signed-off-by: Shirish S <shirish.s@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Register ISR hnadler on the new interrupt.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This will return back MAX cursor size for given ASIC.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Jordan Lazare <Jordan.Lazare@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jordan Lazare <Jordan.Lazare@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This wires DCE12 support into DC and enables it.
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Reviewed-by: Jordan Lazare <Jordan.Lazare@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Modify amdgpu_dm_atomic_comit to implement
atomic_comit_tail hook.
Unify Buffer objects allocation and dealocation
for surface updates and page flips.
Simplify wait for fences and target_vbank logic
for non blockiing commit.
Remove hacky update surface to page flip synchronization
we had and rely on atomic framework synchronization logic.
v2:
Add state->allow_modeset as indicator of page flip call.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
No need to assign them dynamically. This is much more readable.
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Reviewed-by: Jordan Lazare <Jordan.Lazare@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Switch from VUPDATE to VBLANK.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
dc_target does not fit well into DRM framework so removed it.
This will prevent the driver from leveraging the pipe-split
code for tiled displays, so will have to be handled at a higher
level. Most places that used dc_target now directly use dc_stream
instead.
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Problem: MST topology discovery was started before fbdev
init causing kernel panic on boot.
Fix: Defer MST to late init hook, after fbdev strucutres
already initilized.
v2: Add explicit late_init hook function in DAL.
v3:
Change signature of detect_mst_link_for_all_connectors
from void* to drm_dev*
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
v2: agd: squash in dm fix, rebase
Signed-off-by: Jordan Lazare <Jordan.Lazare@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Retain a dc_sink pointer until a new physical pointer
arrives in case of new display connected.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Acked-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Some functions like dm_bandwidth_update are placeholders
for future implementations. Right now, they simply
print messages like "DM_NOT_IMPL: dm_bandwidth_update".
This message even though informational, sometimes can
create some confusion for users.
Since these functions are only skeletons for future, we
simply remove the rather un-unecessary messages from these
functions.
Signed-off-by: Arindam Nath <arindam.nath@amd.com>
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>
Now that the mc_access functions are gone, we no longer
need separate structs for all the different dce families
in dm.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Supported DCE versions: 8.0, 10.0, 11.0, 11.2
v2: rebase against 4.11
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
