DG1 apparently needs the combo PLL fractional divider w/a
with 38.4 MHz refclk as well. This isn't listed in bspec, but
looking at the hsd it looks like it was possibly just missed
due to no one having a DG1 around at the time.
This gives us slightly more accurate clocks on DG1.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250401163752.6412-2-ville.syrjala@linux.intel.com
Reviewed-by: Imre Deak <imre.deak@intel.com>
Going forward, struct intel_display is the main display device data
pointer. Convert the i9xx_wm.h interface to struct intel_display.
With this, we can make intel_wm.c independent of i915_drv.h.
v2: Also remove i915_drv.h, fix commit message
Reviewed-by: Suraj Kandpal <suraj.kandpal@intel.com>
Link: https://lore.kernel.org/r/3e30634d85c0e0aac9c95f9a2f928131ba400271.1744119460.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Unify the naming of the data and clock lane timing parameters, and
simplify their bounds checks. Drop the debug messages on out of bounds
parameters as excessive.
Clarify the comment while at it.
Cc: William Tseng <william.tseng@intel.com>
Reviewed-by: William Tseng <william.tseng@intel.com>
Tested-by: William Tseng <william.tseng@intel.com>
Link: https://lore.kernel.org/r/d1a75ae7b9d93a0b50976b5de45ba2ca798991ad.1743682608.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
The history of why the max of VBT clock and data lane prepare timing
parameter is used for both instead of each individually is
unknown. Separate them to follow what the Windows driver does.
Cc; William Tseng <william.tseng@intel.com>
Reviewed-by: William Tseng <william.tseng@intel.com>
Tested-by: William Tseng <william.tseng@intel.com>
Link: https://lore.kernel.org/r/079a26d0aae79f299aee0397dad2d6519cd55071.1743682608.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
intel_dp::mst.active_links actually indicates the number of MST streams,
not the number of MST links (one MST link carrying one or more MST
streams), rename the field accordingly.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250404150310.1156696-7-imre.deak@intel.com
It's not clear which encoder intel_dp_mst_encoder_active_links() refers
to (primary/stream), but there is also no reason to call the queried
property an encoder property; remove encoder from the name. Also it's
the number of MST streams being queried, vs. the number of MST links
(there is one MST link carrying one or more MST streams), so rename link
to stream as well.
While at it pass intel_dp to the function, which is more logical and
makes it easier to re-use the function later (without the need to get
the digital port pointer).
Also move the function earlier, next to the related ones.
Suggested-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250404150310.1156696-5-imre.deak@intel.com
There doesn't seem to be a reason to assert for a non-negative stream
counter in intel_dp_check_mst_status() in particular, remove it. There
is now an equivalent assert in intel_dp_mst_dec_active_streams().
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250404150310.1156696-4-imre.deak@intel.com
Add helpers to increment/decrement the active MST stream count, instead
of open-coding these.
In mst_stream_pre_enable(), the increment will happen earlier, this is
ok, since nothing depends on the counter between the two points.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250404150310.1156696-3-imre.deak@intel.com
The intel_dp::link_trained flag indicates whether the link is active,
regardless of whether the link training passed or failed. For clarity
rename the flag to 'active'. While at it move the flag under
intel_dp::link.
Suggested-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250404150310.1156696-2-imre.deak@intel.com
If there are no changes to intel_crtc_can_enable_sagv() there
is no need to do all the sagv bw_state recomputation.
The only slight caveat here is hw state takeover where we
initially disable SAGV, and want it to get re-enabled once
we've determined that it's safe to do so. That can now be
achieved by having intel_crtc_can_enable_sagv() reject SAGV
as long as the crtc_state->inherited flag is set. Once the
flag gets cleared (during initial commit for inactive pipes,
during the first userspace commit for active pipes), we
will naturally recompute all the sagv related state.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-14-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Now that all the sagv computation has been moved from the
skl+ watermark code into intel_bw_atomic_check() there is
no point in calling intel_bw_modeset_checks() before the
wm computation. Hide it within intel_bw_atomic_check().
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-13-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
The only thing between the current intel_bw_check_sagv_mask() call
site and intel_bw_atomic_check() is skl_wm_add_affected_planes()
which no longer depends on the sagv mask, so we can make life
a lot less confusing by calling intel_bw_check_sagv_mask() from
intel_bw_atomic_check() instead.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-12-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Move the bw_state->pipe_sagv_reject computation into intel_bw.c
where it belongs.
Previously we had a complicated dance between watermarks and
sagv which required this to be computed earlier, but that was
changed in commit 5e8146251f7b ("extract intel_bw_check_sagv_mask()")
which allows the whole thing to be cleaned up quite a bit.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-11-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
I want to use the crtc_state->inherited flag to clean up some
of the early SAGV handling. To make that work nicely I need to
flag even the inactive crtcs as "inherited".
Since we can't expect user space to perform any real commits
on inactive crtcs we'll clear the flag already during
initial_commit().
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-8-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Currently intel_bw_calc_min_cdclk() always adds the bw_state
to the atomic state. Not only does it result in potentially
redundant work later, it's also currently causing unwanted cdclk
changes during driver load.
Check if the dbuf bw is actually changing before we decide to
pull in the bw state.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-6-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
intel_bw_crtc_min_cdclk() only depends on the pipe data rate,
which we already have stashed in bw_state->data_rate[]. So
stashing the resulting min_cdclk[] as well is redundant. Get
rid of it.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250326162544.3642-2-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
According to the DP spec TPS4 is mandatory for HBR3. We have
however seen some broken eDP sinks that violate this and
declare support for HBR3 without TPS4 support.
At least in the case of the icl Dell XPS 13 7390 this results
in an unstable output.
Reject HBR3 when TPS4 supports is unavailable on the sink.
v2: Leave breadcrumbs in dmesg to avoid head scratching (Jani)
Cc: stable@vger.kernel.org
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/5969
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250306210740.11886-1-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
We are seeing timeouts in opening CRC fd when testing on setup where DP
Panel Replay can be enabled. Fix these by checking if CRC is enabled for DP
Panel Replay as well.
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://lore.kernel.org/r/20250331090747.2964028-1-jouni.hogander@intel.com
Now that we have intel_ggtt_read_entry() we can easily read out the
first PTE of the BIOS FB and verify that it looks correct. We'll
also use the extracted dma address to figure out where in stolen
the FB lives (so far we've just assumed that it sits at offset 0,
and in practice that does seem to be true, but better safe than
sorry).
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250313140838.29742-7-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
The BIOS FB takeover code wants to read out the PTEs (or at least
one of them) to figure out where the FB is located in memory.
Currently we only do that for systems with LMEMBAR, and we've
open coded the PTE decoding in the display code. Introduce a more
proper abstract interface (intel_ggtt_read_entry()) for this purpose,
and implement it for all platforms.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250313140838.29742-5-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
i915 wants to read out the PTE(s) populated by the BIOS/GOP
to verify that the framebuffer is in the correct location.
Introduce intel_gmch_gtt_read_entry() that reads out the
PTE and decodes it to a somewhat abstract form. For now
we just return the dma_addr, present bit, and local memory
bit. I didn't bother with the snoop bit/etc.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250313140838.29742-4-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Usually I'd argue hardcoding values is the wrong thing to do, but in
this case, GVT looking deep into the guts of the DPLL manager for the
reference clocks is worse. This is done for BDW and BXT only, and there
shouldn't be any reason to try to be so dynamic about it.
This helps reduce the direct pokes at display guts from non-display
code.
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Zhenyu Wang <zhenyuw.linux@gmail.com>
Link: https://lore.kernel.org/r/20250321125114.750062-1-jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Forward declare struct drm_printer instead of including drm/drm_print.h,
as we only need the pointer. Turns out quite a few places depend on this
include implicitly. Make them explicit.
Some of the includes are just stale and unnecessary. Group the forward
declarations together while at it.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20250326115452.2090275-1-jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Update GVT-g MAINTAINERS entry to reflect the current status of
maintenance and repositories.
Cc: Dave Airlie <airlied@gmail.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Simona Vetter <simona.vetter@ffwll.ch>
Cc: Tvrtko Ursulin <tursulin@ursulin.net>
Cc: Zhenyu Wang <zhenyuw.linux@gmail.com>
Cc: Zhi Wang <zhi.wang.linux@gmail.com>
Acked-by: Zhi Wang <zhi.wang.linux@gmail.com>
Acked-by: Zhenyu Wang <zhenyuw.linux@gmail.com>
Acked-by: Simona Vetter <simona.vetter@ffwll.ch>
Link: https://lore.kernel.org/r/20250227093805.2217658-1-jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Initializing const char opregion_signature[16] = OPREGION_SIGNATURE
(which is "IntelGraphicsMem") drops the NUL termination of the
string. This is intentional, but the compiler doesn't know this.
Switch to initializing header->signature directly from the string
litaral, with sizeof destination rather than source. We don't treat the
signature as a string other than for initialization; it's really just a
blob of binary data.
Add a static assert for good measure to cross-check the sizes.
Reported-by: Kees Cook <kees@kernel.org>
Closes: https://lore.kernel.org/r/20250310222355.work.417-kees@kernel.org
Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13934
Tested-by: Nicolas Chauvet <kwizart@gmail.com>
Tested-by: Damian Tometzki <damian@riscv-rocks.de>
Cc: stable@vger.kernel.org
Reviewed-by: Zhenyu Wang <zhenyuw.linux@gmail.com>
Link: https://lore.kernel.org/r/20250327124739.2609656-1-jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Instead of logging the dram type in the per version/platform function,
do it in the generic one. This fixes a few discrepancies depending on
the platform:
- There was no DRAM type logging for graphics version 12 and
above
- For graphics version 11, it would log the DRAM type in
skl_get_dram_info(), but could possibly override it later
without any log in icl_pcode_read_mem_global_info()
For bxt_get_dram_info(), there's no need to log the type for each dimm,
as the drm_WARN_ON() already covers the case they are not all the same.
This maintains the behavior of skl_get_dram_info() that would log the
DRAM type even on failures.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20250324-dram-type-v1-2-bf60ef33ac01@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Some new dram types were added without adding the corresponding string
conversion, probably because it's not being used by recent platforms.
Add them, together with a BUILD_BUG_ON() to ensure it keeps in sync, in
preparation to make use of them in recent platforms.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20250324-dram-type-v1-1-bf60ef33ac01@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
