We have at least two options for how to do the
TRANS_PUSH_SEND + commit completion signalling
with the DSB:
Option A)
1. trigger TRANS_PUSH_SEND
2. wait for "safe window"
3. signal the interrupt
In this cases step 2 should not do anything if we were already
between vmin and vmax decision boundaries. Otherwise we'll wait
until the next start of the vblank period.
Option B)
1. wait for "safe window"
2. trigger TRANS_PUSH_SEND
3. signal the interrupt
This option is perhaps a bit less racy, but if we do somehow
screw up and the wait is a nop but the push gets deferred
until the next frame then we'll end up completing the commit
a frame too early.
So for now I'm leaning towards option A since losing the race
won't have any drastic consequences. To deal with the race we
can give the DSB a bit more time to start step 2 before the
hardware has started the vblank termination properly. Often
times it seems to be fast enough to make it in time even without
any extra vblank delay (the push is issued somewhere within a
scanline and it latches on the next scanline).
v2: Use intel_vrr_possible() to determine if we need some
vblank delay (also avoids adding it for DSI which doens't
actually program the transcoder registers correctly for it)
Cc: Paz Zcharya <pazz@chromium.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250116201637.22486-8-ville.syrjala@linux.intel.com
Currently we are forcing full modeset if Panel Replay mode is changed. This
is not necessary as long as we are not changing sink PANEL REPLAY ENABLE
bit in PANEL REPLAY ENABLE AND CONFIGURATION 1 register. This can be
achieved by entering Panel Replay inactive mode (Live Frame mode) when
Panel Replay is disabled and keep PANEL REPLAY ENABLE bit in PANEL REPLAY
ENABLE AND CONFIGURATION 1 enabled always if panel is just supporting Panel
Replay.
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250109103532.2093356-5-jouni.hogander@intel.com
Looks like CCS + async flips has been a thing for a while now.
Enable this for TGL+ render compression modifiers.
Note that we can't update AUX_DIST during async flips we must
check to make sure it remains unchanged.
We also can't do clear color. Supposedly there was some attempt
to make it work, but apparently the issues only got ironed out
in MTL. For now we'll not worry about it and refuse async flips
with clear color modifiers.
Bspec claims that media compression doesn't support async flips.
Based on a quick test it does seem to work to some degree, but
perhaps it has issues as well. Let's trust the spec here and
continue to refuse async flips + media compression.
Bspec: 49250,49251,49252,49253
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241009182207.22900-2-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Pull the vblank delay computation into a separate function.
We'll need more logic here soon and we don't want to pollute
intel_crtc_compute_config() with low level details.
We'll use HAS_DSB() to determine if any delay might be required
or not because delayed vblank only really exists for the
purposes of the DSB. It also doesn't event exists on any pre-tgl
platforms, which also don't have DSB. I was midly tempted
to check for the enable_dsb modparam here actually, but as
that can be changed dynamically via debugfs we'd need to either
reconfigure it on the fly or force a modeset. Neither will happen
currently, so we'll just assume DSB may be used of the platform
supports it.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241210211007.5976-2-ville.syrjala@linux.intel.com
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
At the moment dsc_split represents whether the dsc splitter is used
or not. With 3 DSC engines, the splitter can split into two streams
or three streams.
Instead of representing the splitter's state, it is more effective to
represent the number of DSC streams per pipe.
Replace the `dsc.dsc_split` member with `dsc.num_streams` to indicate the
number of DSC streams used per pipe. This change will implicitly
convey the splitter's operation mode.
v2: Avoid new enum for dsc split. (Suraj)
v3:
-Replace dsc_split with num_stream. (Suraj)
-Avoid extra parentheses. (Jani)
v4: Set num_streams to 1, if VDSC_JOINER not set while readout.
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Reviewed-by: Suraj Kandpal <suraj.kandpal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241030041036.1238006-3-ankit.k.nautiyal@intel.com
Changes in Dynamic Range and Mastering infoframe
should not trigger a full modeset. Therefore, allow
fastset. DP SDP programming is already hooked up in the
fastset flow but HDMI AVI infoframe update is not, add it.
Any other infoframe that can be fastset should be added to
the helper intel_hdmi_fastset_infoframes().
v3:
- Create a wrapper intel_ddi_update_pipe_hdmi to stick to
uniform naming (Jani)
- Do not disable HDMI AVI infoframe if already disabled (Uma)
v2:
- Update HDMI AVI infoframe during fastset.
Signed-off-by: Chaitanya Kumar Borah <chaitanya.kumar.borah@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Animesh Manna <animesh.manna@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241023044122.3889137-1-chaitanya.kumar.borah@intel.com
Panel Replay VSC SDP not getting sent when VRR is enabled
and W1 and W2 are 0. So Program Set Context Latency in
TRANS_SET_CONTEXT_LATENCY register to at least a value of 1.
The same is applicable for PSR1/PSR2 as well.
HSD: 14015406119
v1: Initial version.
v2: Update timings stored in adjusted_mode struct. [Ville]
v3: Add WA in compute_config(). [Ville]
v4:
- Add DISPLAY_VER() check and improve code comment. [Rodrigo]
- Introduce centralized intel_crtc_vblank_delay(). [Ville]
v5: Move to crtc_compute_config(). [Ville]
v6: Restrict DISPLAY_VER till 14. [Mitul]
v7:
- Corrected code-comment. [Mitul]
- dev_priv local variable removed. [Jani]
v8: Introduce late_compute_config() which will take care late
vblank-delay adjustment. [Ville]
v9: Implementation simplified and split into multiple patches.
v10:
- Split vrr changes and use struct intel_display in DISPLAY_VER(). [Ankit]
- Use for_each_new_intel_connector_in_state(). [Jani]
v11: Remove loop and use flipline instead of vrr.enable flag. [Ville]
v12:
- Use intel_Vrr_possible helper.
- Correct flag check for flipline.
v13:
- Refactor workaround [Jonathan]
- Drop the comment around woraround number. [Ville]
Signed-off-by: Animesh Manna <animesh.manna@intel.com>
Signed-off-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
[vsyrjala: Make adjusted_modeg const, and drop redundant parens]
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241010040503.1795399-4-mitulkumar.ajitkumar.golani@intel.com
Push regular plane/color management updates to the DSB,
if other constraints allow it.
The first part of the sequence will go as follows:
- CPU will kick off DSB0 immediately
- DSB0 writes double bufferd non-arming registers
- DSB0 evades the vblank
- DSB0 writes double buffered arming registers
If no color management updates is needed we follow that up with:
- DSB0 waits for the undelayed vblank
- DSB0 waits for the delayed vblank (usec wait)
- DSB0 emits an interrupt which will cause the CPU to complete the commit
If color management update is needed:
- DSB0 will start DSB1 with wait for undelayed vblank
- DSB0 will in parallel perform the force DEwake tricks
- DSB1 writes single buffered LUT registers
- DSB1 waits for the delayed vblank (usec wait)
- DSB1 emits an interrupt which will cause the CPU to complete the commit
With this sequence we don't need to increase the vblank delay
to make room for register programming during vblank, which is
a good thing for high refresh rate display. But I'll need to
still think of some way to eliminate VRR commit completion
related races under this scheme.
Stuff that isn't ready for DSB yet:
- modesets (potentially we could do
at least the plane enabling via DSB)
- fastsets
- VRR
- PSR
- scalers
- async flips
Reviewed-by: Animesh Manna <animesh.manna@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240930170415.23841-14-ville.syrjala@linux.intel.com
We need to be able to do both MMIO and DSB based pipe/plane
programming. To that end plumb the 'dsb' all way from the top
into the plane commit hooks.
The compiler appears smart enough to combine the branches from
all the back-to-back register writes into a single branch.
So the generated asm ends up looking more or less like this:
plane_hook()
{
if (dsb) {
intel_dsb_reg_write();
intel_dsb_reg_write();
...
} else {
intel_de_write_fw();
intel_de_write_fw();
...
}
}
which seems like a reasonably efficient way to do this.
An alternative I was also considering is some kind of closure
(register write function + display vs. dsb pointer passed to it).
That does result is smaller code as there are no branches anymore,
but having each register access go via function pointer sounds
less efficient.
Not that I actually measured the overhead of either approach yet.
Also the reg_rw tracepoint seems to be making a huge mess of the
generated code for the mmio path. And additionally there's some
kind of IS_GSI_REG() hack in __raw_uncore_read() which ends up
generating a pointless branch for every mmio register access.
So looks like there might be quite a bit of room for improvement
in the mmio path still.
Reviewed-by: Animesh Manna <animesh.manna@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240930170415.23841-12-ville.syrjala@linux.intel.com
Once we start using DSB for plane updates we'll need to defer
generating the DSB buffer until the clear color has been
read out. So we need to move at some of the DSB stuff into
commit_tail(). That is perhaps a better place for it anyway
as the ioctl thread can move on immediately without spending
time building the DSB commands.
We always have the MMIO fallback (in case the DSB buffer
allocation fails), so there's no real reason to keep any
of this in the synchronous part of the ioctl.
Because the DSB LUT programming doesn't depend on the plane
clear color we can still do that part before waiting for
fences/etc. which should help paralleize things a bit more.
The DSB plane programming will need to happen after those
however as that depends on the clear color.
Reviewed-by: Animesh Manna <animesh.manna@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240930170415.23841-4-ville.syrjala@linux.intel.com
