Normally we could be in a deep PkgC state all the way up to the
point when DSB starts its execution at the transcoders undelayed
vblank. The DSB will then have to wait for the hardware to
wake up before it can execute anything. This will waste a huge
chunk of the vblank time just waiting, and risks the DSB execution
spilling into the vertical active period. That will be very bad,
especially when programming the LUTs as the anti-collision logic
will cause DSB to corrupt LUT writes during vertical active.
To avoid these problems we can instruct the DSB to pre-wake the
display engine on a specific scanline so that everything will
be 100% ready to go when we hit the transcoder's undelayed vblank.
One annoyance is that the scanline is specified as just that,
a single scanline. So if we happen to start the DSB execution
after passing said scanline no DEwake will happen and we may drop
back into some PkgC state before reaching the transcoder's undelayed
vblank. To prevent that we'll use the "force DEwake" bit to manually
force the display engine to stay awake. We'll then have to clear
the force bit again after the DSB is done (the force bit remains
effective even when the DSB is otherwise disabled).
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606191504.18099-18-ville.syrjala@linux.intel.com
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
We want to start the DSB execution from the transcoder's undelayed
vblank, so in order to guarantee atomicity with the all the other
mmio register writes we need to evade both vblanks.
Note that currently we don't add any vblank delay, so this is
effectively a nop. But in the future when we start to program
double buffered registers from the DSB we'll need to delay the
pipe's vblank to provide the register programming "window2"
for the DSB.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606191504.18099-15-ville.syrjala@linux.intel.com
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Loading LUTs with the DSB outside of vblank doesn't really
work due to the palette anti-collision logic. Apparently the
DSB register writes don't get stalled like CPU mmio writes
do and instead we end up corrupting the LUT entries. Disabling
the anti-collision logic would allow us to successfully load
the LUT outside of vblank, but presumably that risks the LUT
reads from the scanout (temporarily) getting corrupted data
from the LUT instead.
The anti-collision logic isn't active during vblank so that
is when we can successfully load the LUT with the DSB. That is
what we want to do anyway to avoid tearing.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606191504.18099-13-ville.syrjala@linux.intel.com
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Starting with Xe_LP+, GFX_MSTR_IRQ contains status bits that have W1C
behavior. If we do not properly reset them, we would miss delivery of
interrupts if a pending bit is set when enabling IRQs.
As an example, the display part of our probe routine contains paths
where we wait for vblank interrupts. If a display interrupt was already
pending when enabling IRQs, we would time out waiting for the vblank.
Avoid the potential issue by clearing GFX_MSTR_IRQ as part of the IRQ
reset.
v2:
- Move logic from gen11_gt_irq_reset() to dg1_irq_reset(). (Matt)
BSpec: 50875, 54028
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920195351.59421-2-gustavo.sousa@intel.com
We have the same h/vsync_end vs. h/vtotal quirk in the VBT parser
that was also present in EDID parser. Adjust the VBT parser the
same way as was done for hte EDID parser to fixup h/vsync_end
instead of h/vtotal. While I'm not currently aware of any machines
that need this for the VBT it seems prudent to keep both parsers
in sync.
And while at it let's add some debugs here as well. A bit
lackluster but didn't feel like plumbing the connector all
the way down at this time.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920211934.14920-2-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Introduce correspondent definitions for choosing between CD2X CDCLK
and PLL CDCLK as a source. All the entries in cdclk table for xe2lpd are
defined with PLL CDCLK as source, so simply set it. Also
skl_cdclk_decimal() shouldn't be set in CDCLK_CTL anymore, so skip it
for display version 20 and above.
v2:
- Remove unneeded comment and use REG_BIT() (Matt Roper)
- Rename CDCLK_SOURCE_SEL_CDCLK_PLL() to MDCLK_SOURCE_SEL_CDCLK_PLL
to match spec (Lucas)
Signed-off-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230919192128.2045154-22-lucas.demarchi@intel.com
Add a new CDCLK table for Lunar Lake.
v2:
- Remove mdclk from the table as it's not needed (Matt Roper)
- Update waveform values to the latest from spec (Matt Roper)
- Rename functions and calculation to match by pixel rate (Lucas)
v3: Keep only the table: as far as intel_pixel_rate_to_cdclk()
is concerned, the minimum cdclk should still be half the pixel
rate on Xe2 (bspec 68858:
"Pipe maximum pixel rate = 2 * CDCLK frequency * Pipe Ratio")
(Matt Roper)
Bspec: 68861, 68858
Signed-off-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230919192128.2045154-19-lucas.demarchi@intel.com
Xe2_LPD doesn't have south display engine on a PCH, it's actually
on the SoC die (while north display engine is on compute die). As
such it makes no sense to go through the PCI devices looking for
an ISA bridge. The approach used by BXT/GLK can't be used here since
leaving it with PCH_NONE would mean taking the wrong code paths.
For the places we currently use a PCH check, it's enough for now to just
check the north display version. Use that to define a fake PCH to be
used across the driver. Eventually these PCH checks may need to be
re-designed as this is already the third platform using/needing a
fake PCH.
v2: Match on display IP version rather than on platform (Matt Roper)
v3: Extend and clarify commit message (Matt Roper / Ville)
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230919192128.2045154-6-lucas.demarchi@intel.com
This reverts commit 959fb1a686.
Looks like the core MST code might not call i2c_adapter() for us
in time, and thus creating the ddc symlink will fail. This will
in fact fail the entire connector registration, but the MST code
doesn't really seem to care about that and blindly plows ahead.
All we may get in the logs is a nearly back to back
register+unregister debug messages:
[drm:drm_dp_mst_connector_late_register [drm_display_helper]] registering DPMST remote bus for card0-DP-7
[drm:intel_dp_hpd_pulse [i915]] DPRX ESI: 42 00 00 02
[drm:drm_dp_mst_connector_early_unregister [drm_display_helper]] unregistering DPMST remote bus for card0-DP-7
Untangling the initialization order may take some real work,
so let's just revert the ddc symlink addition for now...
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9357
Fixes: 959fb1a686 ("drm/i915/mst: Populate connector->ddc")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230919095659.10742-1-ville.syrjala@linux.intel.com
Acked-by: Jani Nikula <jani.nikula@intel.com>
Tested-by: Karthik B S <karthik.b.s@intel.com>
Implement low refresh rate (LRR) where we change the vblank
length by hand as requested, but otherwise keep the timing
generator running in non-VRR mode (ie. fixed refresh rate).
The panel itself must support VRR for this to work, and
only TGL+ has the double buffred TRANS_VTOTAL.VTOTAL that
we need to make the switch properly. The double buffer
latching happens at the start of transcoders undelayed
vblank. The other thing that we change is
TRANS_VBLANK.VBLANK_END but the hardware entirely ignores
that in DP mode. But I decided to keep writing it anyway
just to avoid more special cases in readout/state check.
v2: Document that TRANS_VBLANK.VBLANK_END is ignored by
the hardware
v3: Reconcile with VRR fastset
Adjust update_lrr flag behaviour
Make sure timings stay within VRR range
v4: Fix up update_m_n vs. update_lrr rebase fail (Manasi)
Drop DOUBLE_BUFFER_VACTIVE define as it's not needed (Manasi)
TODO: Hook LRR into the automatic DRRS downclocking stuff?
Cc: Manasi Navare <navaremanasi@chromium.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230915103800.14218-1-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <navaremanasi@chromium.org>
Imre Deak
Ville Syrjälä
Jani Nikula
Gustavo Sousa
Tvrtko Ursulin
Balasubramani Vivekanandan
Stanislav Lisovskiy
Matt Roper
Ravi Kumar Vodapalli
Lucas De Marchi
Juha-Pekka Heikkilä
Luca Coelho
Clint Taylor