drm/doc: fix duplicate declaration warning

We have duplicate kernel-doc directives for the same struct, leading to:

/home/jani/src/linux/Documentation/gpu/driver-uapi.rst:2279: WARNING: Duplicate C declaration, also defined at rfc/i915_scheduler:3.
Declaration is '.. c:struct:: i915_context_engines_parallel_submit'.

Use the Sphinx C domain namespace for the rfc document to fix this.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Luca Coelho <luciano.coelho@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230621123156.14907-1-jani.nikula@intel.com

drm/i915: Prevent needless toggling of DC states during modesets

DC states are disabled / re-enabled around each modeset, which may lead
to a needless off->on->off toggling of the DC_off power well. This has
some overhead as toggling DC states involves running a DMC firmware
handler and also running a periodic firmware handler while DC states are
enabled. The limit of when DC states have a benefit is at 30 FPS (using
DC3co) and below 30 FPS (using DC5/6), where the firmware can actually
disable clocks / power off power wells. Accordingly delay powering off
the DC_off powerwell (which re-enables DC states) by 17 ms at the end of
a modeset to avoid the above overhead at or above 60 FPS.

Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616185104.2502003-4-imre.deak@intel.com

drm/i915: Add way to specify the power-off delay of a display power domain

Add support for specifying a delay different than the current 100 ms
default for powering off a display power domain. This is needed by the
next patch which delays re-enabling DC states during modesets to avoid
the off->on->off toggling overhead of the DC_off power well, but does
this using a < 100 ms delay for a better utilization of DC power saving
states.

Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616185104.2502003-3-imre.deak@intel.com

drm/i915: Remove redundant forward declarations from display power headers

Remove the forward declarations for enums and structs from display power
header files that aren't used in prototypes.

Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616185104.2502003-2-imre.deak@intel.com

drm/i915: Add missing forward declarations/includes to display power headers

Add the seq_file struct forward declaration to intel_display_power.h
fixing the build error below. While at it add the rest of missing
forward declarations/includes to the display power header files.

In file included from <command-line>:
./../drivers/gpu/drm/i915/display/intel_display_power.h:255:70: error: 'struct seq_file'
declared inside parameter list will not be visible outside of this definition or declaration [-Werror]
  255 | void intel_display_power_debug(struct drm_i915_private *i915, struct seq_file *m);
      |                                                                      ^~~~~~~~

Closes: https://lore.kernel.org/intel-gfx/89adc1ac-25a0-6eb6-4cc9-ab6cc8d49730@infradead.org/
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Tested-by: Randy Dunlap <rdunlap@infradead.org> # build-tested
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616185104.2502003-1-imre.deak@intel.com

drm/i915/hdcp: Add a debug statement at hdcp2 capability check

Add a debug statement at hdcp2 capability check which indicates if
GSC CS is causing hdcp2 incapability

--v2
-correcttypo in commit header

--v3
-correct the other typo in commit header [Jani]

Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Chaitanya Kumar Borah <chaitanya.kumar.borah@intel.com>
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230621085453.1996166-1-suraj.kandpal@intel.com

drm/i915/hdcp: Assign correct hdcp content type

Currently hdcp->content_type is being assigned the content_type field
in drm_connector_state which is wrong and instead it needs to be
assigned hdcp_content_type field from drm_connector_state

Fixes: 4c4279a8d58d ("drm/i915/hdcp: add intel_atomic_state argument to hdcp_enable function")
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Cc: Animesh Manna <animesh.manna@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230622083254.2057102-1-suraj.kandpal@intel.com

drm/i915/psr: Use hw.adjusted mode when calculating io/fast wake times

Encoder compute config is changing hw.adjusted mode. Uapi.adjusted mode
doesn't get updated before psr compute config gets called. This causes io
and fast wake line calculation using adjusted mode containing values before
encoder adjustments. Fix this by using hw.adjusted mode instead of
uapi.adjusted mode.

Cc: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/8475
Fixes: cb42e8ede5b4 ("drm/i915/psr: Use calculated io and fast wake lines")
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230620111745.2870706-1-jouni.hogander@intel.com

drm/i915/mtl: Skip using vbt hdmi_level_shifter selection on MTL

The hdmi_level_shifter part of General Bytes definition in VBT, which was
used for choosing different levels on earlier platforms is now a hidden
optin and shows the default value of 0. The level shifter is now to be
deduced from hdmi_default_entry in  intel_ddi_buf_trans for each phy.

Skip providing the default hw provided value to force driver to choose hdmi
default entry.

Bspec: 20124
Cc: Khaled Almahallawy <khaled.almahallawy@intel.com>
Cc: Lee Shawn C <shawn.c.lee@intel.com>
Cc: Clint Taylor <Clinton.A.Taylor@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Reviewed-by: Clint Taylor <Clinton.A.Taylor@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616210028.1601533-1-radhakrishna.sripada@intel.com

drm/i915/mtl: Fix SSC selection for MPLLA

Driver does not clear the default SSC for MPLLA. This causes link training
failure when trying to use 10G and 20G rates. Fix the behaviour and enable
ssc only when we really want.

Fixes: 237e7be0bf57 ("drm/i915/mtl: For DP2.0 10G and 20G rates use MPLLA")
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Clint Taylor <Clinton.A.Taylor@intel.com>
Cc: Khaled Almahallawy <khaled.almahallawy@intel.com>
Cc: Arun R Murthy <arun.r.murthy@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Tested-by: Khaled Almahallawy <khaled.almahallawy@intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616043950.1576836-1-radhakrishna.sripada@intel.com

drm/i915/mtl: Add new vswing table for C20 phy to support DP 1.4

Add vswing table to support DP 1.4 for C20 phy.

v2: rename mtl_c10_trans
v3: add default_entry for mtl_c20_trans_dp14
v4: rename mtl_cx0_trans_dp14

Bspec: 74104
Signed-off-by: Lee Shawn C <shawn.c.lee@intel.com>
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Clint Taylor <Clinton.A.Taylor@intel.com>
Cc: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609002038.11516-1-shawn.c.lee@intel.com

drm/i915/mtl: update DP 2.0 vswing table for C20 phy

Update preset 15 setting to align the latest bspec value.

Bspec: 74104
Signed-off-by: Lee Shawn C <shawn.c.lee@intel.com>
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Clint Taylor <Clinton.A.Taylor@intel.com>
Cc: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Cc: Uma Shankar <uma.shankar@intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606104238.31953-1-shawn.c.lee@intel.com

drm/i915: Convert HSW/BDW to use port_mask for DDI probe

Make HSW/BDW use port_mask for output probing as well.
To achieve that the strap checks are moved into
intel_ddi_init() itself. Or should we move them to the
runtime port_mask init instead? Maybe not since the hardware
is still there, just not connected to anything.

v2: Account for DDI-E in strap detection
    Keep to the old CRT->DDI init order

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616140820.11726-8-ville.syrjala@linux.intel.com

drm/i915: Init DDI outputs based on port_mask on skl+

Instead of listing every platform's possible DDI outputs
in intel_setup_outputs() just loop over the new port_mask
to achieve the same thing.

HSW/BDW were left as is since they still look at the straps
as well.

DSI is still a mess. For now just check for the relevant
platforms explicitly.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616140820.11726-7-ville.syrjala@linux.intel.com

drm/i915: Beef up SDVO/HDMI port checks

The SDVO code already warns when the port in question doesn't
actually support SDVO. Let's make that also bail the encoder
registration like the generic assert_port_valid() we added.

And add a similar thing for g4x HDMI, mainly because on g4x
itsefl port D only supports DP but not SDVO/HDMI. For the
other platforms the generic port_mask check should actually
be sufficient, but since we're here might as well list the
ports.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616140820.11726-6-ville.syrjala@linux.intel.com

drm/i915: Assert that the port being initialized is valid

Sprinkle some asserts to catch any mishaps in the port_mask
vs. output init.

For DDI/DP/HDMI/SDVO I decided that we want to bail out for
an invalid port since those are the encoder types where
we might want consider driving the whole thing from the VBT
child device list, and bogus VBTs could be a real issue
(if for no other reason than the i915.vbt_firmware).

For DVO and HSW/BDW CRT port I just threw the assert in
there for good measure.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616140820.11726-5-ville.syrjala@linux.intel.com

drm/i915: Assert that device info bitmasks have enough bits

Sprinkle in some BUILD_BUG_ON()s to make sure some of
the bitmasks used in the device info have enough bits.

Do we have a better place for this sort of stuff?

v2: Relocate to the new place

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616140820.11726-4-ville.syrjala@linux.intel.com

drm/i915: Introduce device info port_mask

Declare the available DVO/SDVO/HDMI/DP/DDI ports in the
device info. The other outputs (LVDS/TV/DSI/VGA) are left
out since for most of them we don't consider them as "ports".

DSI we should probably perhaps include somehow in the device
info. Just not sure how. Or we just introduce a HAS_DSI() and
call it a day?

TODO: figure out what to do about the subplatform stuff. Would
      it be better to declare those directly with a different
      device info or not? Also not sure the icl port-f stuff
      matters even. Bspec claims there are icl SKUs with far
      less ports than that and we don't seem to check for those
      either?

v2: Fix TC5 vs. TC6 mixup on TGL (Jani)
    Drop DDI C for now on TGL, and add a FIXME (Jani)

Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616140820.11726-3-ville.syrjala@linux.intel.com

drm/i915: Remove bogus DDI-F from hsw/bdw output init

HSW/BDW don't have DDI-F so don't go looking for one.

Seems to have been accidentally left behind when the
skl+ stuff got split out in commit 097d9e902068
("drm/i915/display: remove strap checks from gen 9").

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230616140820.11726-2-ville.syrjala@linux.intel.com

drm/i915/adlp+: Allow DC states along with PW2 only for PWB functionality

A recent bspec update added a restriction on when DC states can be enabled:

[Before enabling DC states:]

"""
PG2 can be kept enabled only because PGB requires PG2.
Do not use PG2 functions, such as type-C DDIs.

DMC will dynamically control PG1, PGA, PG2, PGB.
"""

Accordingly prevent DC states if PW2 (aka PG2) is enabled for any other
functionality.

Bpsec: 49193

Fixes: 88c487938414 ("drm/i915: Use separate "DC off" power well for ADL-P and DG2")
Reported-by: Kai Vehmanen <kai.vehmanen@intel.com>
Tested-by: Ambica Pramod <ambica.pramod@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606172822.1891897-1-imre.deak@intel.com

drm/i915/psr: Re-enable PSR1 on hsw/bdw

All known issues fixed now, so re-enable PSR1 on hsw/bdw.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-14-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Allow PSR with sprite enabled on hsw/bdw

Can't see why we'd want the sprite blocking PSR entry.
Mask it out.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-13-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Don't skip both TP1 and TP2/3 on hsw/bdw

WA 0479 says: "Do not skip both TP1 and TP2/TP3". Let's just
stick the minimum 100us TP2/3 time in there to avoid that.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-12-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Do no mask display register writes on hsw/bdw

hsw/bdw lack the pipe register vs. display register distinction
in their PSR masking capabilities. So to keep our CURSURFLIVE
tricks working we need to just unmask all display register writes
on these platforms. The downside being that any display regitster
(eg. even SWF regs) will cause a PSR exit.

Note that WaMaskMMIOWriteForPSR asks us to mask this on bdw, but
that won't work since we need those CURSURFLIVE tricks. Observations
on actual hardware show that this causes one extra PSR exit ~every
10 seconds, which is pretty much irrelevant. I suspect this is
due to the pcode poking at IPS_CTL. Disabling IPS does not stop it
however, so either I'm wrong or pcode pokes at the register
regardless of whether it's actually trying to enable/disable IPS.
Also when the machine is busy (eg. just running 'find /') these
extra PSR exits cease, which again points at pcode or some other
PM entity as being the culprit.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-11-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Implement WaPsrDPRSUnmaskVBlankInSRD:hsw

Bspec asks us to unmask "vblank to registers" in the DPRS unit.

Note that I was unable to observe any change in hardware
behviour due to this bit on HSW. But let's do this anyway
in case it matters in some cases, and the corresponding bit
on BDW is abolutely critical as without it the hardware
won't generate any vblanks whatsoever after PSR exit.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-10-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Implement WaPsrDPAMaskVBlankInSRD:hsw

Implement WaPsrDPAMaskVBlankInSRD:hsw, which makes the hardware
generate the extra vblank between link training and first frame
being transmitted. This is the same thing that's controlled by
TRANS_CHICKEN[21] on skl+ (but due to the funky double buffering
it's effectively always at the rest value after DC5 exit). So
for consistent behaviour we want every platform to generate said
vblank. BDW is already setting this up correctly.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-9-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Restore PSR interrupt handler for HSW

Add the PSR interrupt handling code back for HSW. Looks like
the removal was never completed anyway since the irq setup
code was lest untouched.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-8-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: HSW/BDW have no PSR2

Deal with HSW/BDW in transcoder_has_psr2().

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-7-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Bring back HSW/BDW PSR AUX CH registers/setup

Reintroduce the special PSR AUX CH setup for hsw/bdw. Not all
of it was even removed (BDW AUX data registers were left behind).
Update the code to use REG_BIT() & co. while at it.

v2: Define the SRD_AUX_CTL bits in terms of DP_AUX_CTL bits (Jouni)
    Add a comment explaining the hand rolled DPCD write (Jouni)

Cc: Jouni Högander <jouni.hogander@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-6-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Reintroduce HSW PSR1 registers

Add back hsw'w special SRD/PSR1 registers.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-5-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Wrap PSR1 register with functions

In preparation for re-introducing HSW's different PSR1 register
offeets let's just wrap all the registers into functions.
Avoids having to make the register macros more complex.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-4-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/psr: Fix BDW PSR AUX CH data register offsets

The multiplication got replaced by an addition in some cleanup.
This means we never write the correct data to some of the BDW
PSR data registers and thus we fail to actually wake up the
panel from PSR.

Fixes: 4ab4fa103217 ("drm/i915/psr: Make PSR registers relative to transcoders")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-3-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915: Re-init clock gating on coming out of PC8+

PC8+ clobbers a bunch of displays registers which need to
be restored by hand or else we lost a bunch of workarounds.
The important ones for us are at least CHICKEN_PAR2* and
CHICKEN_PIPESL*.

Curiously at least some CHICKEN_PAR1* registers
are preserved by the hardware/firmware. Unfortunately Bspec
doens't really specify what gets clobbered vs. preserved
so further reverse engieering might be warranted to figure
out the specifics.

Note that PCH_LP_PARTITION_LEVEL_DISABLE is also set by
lpt_init_clock_gating() so the rmw in hsw_disable_pc8()
is now redundant. Remove it.

TODO: I suspect most gt stuff doesn't need this and we should
      finish moving all of them from init_clock_gating() to
      a more appropriate place...

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609141404.12729-2-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/mtl: Cleanup usage of phy lane reset

From PICA message bus we wait for acknowledgment from
read/write commands. In case of an error, we reset the
bus for the next command.

Current implementation ends up resetting message bus twice
in cases where error is not the timeout. Since, we only need
to reset message bus once, let's move reset to corresponding
timeout error and drop the excess reset function calls from
read/write functions.

Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609122130.69794-1-mika.kahola@intel.com

drm/i915/hdcp: Modify hdcp_gsc_message msg sending mechanism

Allocate a multipage object that can be used for input
and output for intel_hdcp_gsc_message so that corruption of
output message can be avoided by the current overwriting method.
Modify intel_gsc_send_sync() to take into account header_out
and addr_out so as to use them to verify the message send status.
Check link for comment and review history
https://patchwork.freedesktop.org/series/118499/

--v2
-Change approach from allocating two objects to just one multipage
object [Daniele]

--v3
-Squash the two patches together [Daniele]
-No need to allocate twice the size of intel_hdcp_gsc_message [Daniele]

Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Cc: Alan Previn <alan.previn.teres.alexis@intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230602105113.1791104-1-suraj.kandpal@intel.com

drm/i915/dsi: Remove weird has_pch_encoder asserts

No idea why the DSI code is feeling the need to assert that
has_pch_encoder must not be set. PCH encoders aren't even a
thing on any platform that has DSI.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-14-ville.syrjala@linux.intel.com

drm/i915/dsi: Grab the crtc from the customary place

The encoder hooks already get passed the crtc state so just
grab the crtc from there instead of from the connector state.
This is generally what everyone else does, so no reason for
icl_dsi.c to be different.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-13-ville.syrjala@linux.intel.com

drm/i915/dsi: Move panel reset+power off to be the last thing

Follow what Windows does and do the DSI panel reset+power off
after everything else.

v2: s/intel_dsi_msleep/msleep/

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-12-ville.syrjala@linux.intel.com

drm/i915/dsi: Respect power_off_delay on icl+

icl+ DSI isn't respecting the panel power_off_delay. Remedy that.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-11-ville.syrjala@linux.intel.com

drm/i915/dsi: Do DSC/scaler disable earlier on icl+

Do the scaler disable in the spot where bspec has specfied it
for TGL+. And also move the DSC disable to match what
intel_ddi.c does.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-10-ville.syrjala@linux.intel.com

drm/i915/dsi: Move most things from .disable() into .post_disable() on icl+

encoder->disable() is supposed to happen before the pipe/transcoder
gets disabled. The icl+ DSI code screwed that up and put most things
(including the transcoder disable itself) into  encoder->disable().
Follow the common rules and hoist most things into the
encoder->post_disable() hook.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-9-ville.syrjala@linux.intel.com

drm/i915/dsi: Implement encoder->shutdown() for icl+

Plug in the encoder->shutdown() hook for icl+ DSI so that
we are guaranteed to respect the power cycle delay during
reboots and whatnot.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-8-ville.syrjala@linux.intel.com

drm/i915/dsi: Respect power cycle delay on icl+

Handle the DSI panel power cycle delay on icl+.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-7-ville.syrjala@linux.intel.com

drm/i915/dsi: Gate DSI clocks earlier

The clock gating step is in the wrong spot compared to the
TGL+ bspec sequence. Move it the right place. Windows also
seems to use the TGL+ order here always.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-6-ville.syrjala@linux.intel.com

drm/i915/dsi: Split icl+ D-PHY vs. DSI timing steps

The programming of the DPHY vs. DSI _TIMING registers are
two separate steps in the TGL+ bspec sequence, with some
other stuff in between. Implement the same split.
Windows also seems follow the bspec TGL+ sequence, even
on ICL/JSL.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-5-ville.syrjala@linux.intel.com

drm/i915/dsi: Print the VBT MIPI sequence delay duration

Help out debugging things by printing out how long the VBT
delay sequence is supposed to wait.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-4-ville.syrjala@linux.intel.com

drm/i915/dsi: Do display on sequence later on icl+

Doing the init OTP and display on DSI sequences back to back
doesn't really make any sense (a single sequence would suffice
then). Move the display on sequence to be done just before
backlight on, which is also what Windows does.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-3-ville.syrjala@linux.intel.com

drm/i915/dsi: Do panel power on + reset deassert earlier on icl+

Looks like we're trying to talk to the DSI panel even before turning
it on, on icl+. Bspec doesn't actually specify when these should be
done, but certainly we need to turn the panel on at least before
talking to it. So let's move the power on + reset deassert steps to
be the first thing we do. This is also what Windows does.

v2: s/intel_dsi_msleep/msleep/

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608203057.23759-2-ville.syrjala@linux.intel.com

drm/i915/gmch: avoid unused variable warning

When CONFIG_PNP is disabled, the mchbar_addr variable is only written but
not read:

drivers/gpu/drm/i915/soc/intel_gmch.c: In function 'intel_alloc_mchbar_resource':
drivers/gpu/drm/i915/soc/intel_gmch.c:41:13: error: variable 'mchbar_addr' set but not used [-Werror=unused-but-set-variable]
   41 |         u64 mchbar_addr;
      |             ^~~~~~~~~~~

No idea why this showed up now, but it's easy to fix by changing the #ifdef to
an IS_ENABLED() check that the compiler can see through.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230612124408.521325-1-arnd@kernel.org

Merge drm/drm-next into drm-intel-next

Sync up with changes from drm-intel-gt-next.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/selftests: add local workqueue for SW fence selftest

Instead of using a global workqueue for the SW fence selftest,
allocate a separate one temporarily only while running the test.

Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/313f4a713053c2b4896ced5b0e9ff456eb85fe57.1686231190.git.jani.nikula@intel.com

drm/i915: add a dedicated workqueue inside drm_i915_private

In order to avoid flush_scheduled_work() usage, add a dedicated
workqueue in the drm_i915_private structure.  In this way, we don't
need to use the system queue anymore.

This change is mostly mechanical and based on Tetsuo's original
patch[1].

v6 by Jani:
- Also create unordered_wq for mock device

Link: https://patchwork.freedesktop.org/series/114608/
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/c816ebe17ef08d363981942a096a586a7658a65e.1686231190.git.jani.nikula@intel.com

drm/i915: use pointer to i915 instead of rpm in wakeref

Currently a pointer to an intel_runtime_pm structure is stored in the
wake reference structures so the runtime data can be accessed.  We can
save the entire device information (drm_i915_private) instead, since
we'll need to reference the new workqueue we'll add in subsequent
patches.

Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ec0eb5149120d04f3d9870d7671ef10103e6fc29.1686231190.git.jani.nikula@intel.com

Merge tag 'drm-intel-gt-next-2023-06-08' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

UAPI Changes:

- I915_GEM_CREATE_EXT_SET_PAT for Mesa on Meteorlake.

Driver Changes:

Fixes/improvements/new stuff:

- Use large rings for compute contexts (Chris Wilson)
- Better logging/debug of unexpected GuC communication issues (Michal Wajdeczko)
- Clear out entire reports after reading if not power of 2 size (Ashutosh Dixit)
- Limit lmem allocation size to succeed on SmallBars (Andrzej Hajda)
- perf/OA capture robustness improvements on DG2 (Umesh Nerlige Ramappa)
- Fix error code in intel_gsc_uc_heci_cmd_submit_nonpriv() (Dan Carpenter)

Future platform enablement:

- Add workaround 14016712196 (Tejas Upadhyay)
- HuC loading for MTL (Daniele Ceraolo Spurio)
- Allow user to set cache at BO creation (Fei Yang)

Miscellaneous:

- Use system include style for drm headers (Jani Nikula)
- Drop legacy CTB definitions (Michal Wajdeczko)
- Turn off the timer to sample frequencies when GT is parked (Ashutosh Dixit)
- Make PMU sample array two-dimensional (Ashutosh Dixit)
- Use the correct error value when kernel_context() fails (Andi Shyti)
- Fix second parameter type of pre-gen8 pte_encode callbacks (Nathan Chancellor)
- Fix parameter in gmch_ggtt_insert_{entries, page}() (Nathan Chancellor)
- Fix size_t format specifier in gsccs_send_message() (Nathan Chancellor)
- Use the fdinfo helper (Tvrtko Ursulin)
- Add some missing error propagation (Tvrtko Ursulin)
- Reduce I915_MAX_GT to 2 (Matt Atwood)
- Rename I915_PMU_MAX_GTS to I915_PMU_MAX_GT (Matt Atwood)
- Remove some obsolete definitions (John Harrison)

Merges:

- Merge drm/drm-next into drm-intel-gt-next (Tvrtko Ursulin)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZIH09fqe5v5yArsu@tursulin-desk

Merge tag 'drm-intel-next-2023-06-05' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

drm/i915 features for v6.5:

Features and functionality:
- Meteorlake (MTL) display enabling (Mika, Radhakrishna, José, Ankit, Clint,
  Gustavo, Imre, Anusha, Juha-Pekka, Matt)
- Allow VRR to be toggled during fastsets (Ville)
- Allow arbitrary refresh rates with VRR eDP panels (Ville)
- Support async flips on linear buffers on display ver 12+  (Arun)
- New debugfs for display clock frequencies (Bhanuprakash)
- Taint kernel when force probing unsupported devices (Jani)
- Expose CRTC CTM property on ILK/SNB/VLV (Ville)

DRM subsystem changes:
- EDID changes to support further conversion to struct drm_edid (Jani)
- Move i915 DSC parameter code to common DRM helpers (Dmitry Baryshkov)

Refactoring and cleanups:
- CSC color refactoring (Ville)
- VRR cleanups (Ville)
- Finish i915 conversion to struct drm_edid (Jani)
- Start high level display driver file (Jani)
- Hotplug refactoring (Ville)
- Misc display refactoring and cleanups (Jani, Ville)
- Use device based logging for state checker warnings (Jani)
- Split out hotplug and display irq handling (Jani)
- Move display device info and probe under display/ (Matt)
- HDCP cleanups (Suraj)
- Use localized warning ignores instead of per file (Jani)
- Remove superfluous enum i915_drm_suspend_mode (Maarten)
- PSR, pfit, scaler and chicken register definition cleanups (Ville)
- Constify pointers to hwmon_channel_info (Krzysztof Kozlowski)
- Replace all non-returning strlcpy with strscpy (Azeem Shaikh)
- Refactor VBT aux channel and DDC pin mapping (Ville)
- Include cleanups (Jani)

Fixes:
- Fix modeset locking issue in DP MST HDCP (Suraj)
- Fix disconnected Type-C/DP-alt disable at probe (Imre)
- Fix HDMI PCON DSC usage and color conversions (Ankit)
- Fix g4x HDMI infoframe/audio transmission port usage (Ville)
- Avoid use-after-free when DP connector init fails (Maarten)
- Fix voltage level for 480 MHz CDCLK (Chaitanya)
- Check HPD live state during eDP probe (Ville)
- Fix active port PLL selection for secondary MST streams (Imre)
- Check pipe source size when using SKL+ scalers (Ville)
- Fix MIPI DSI sleep sequences (Hans de Goede)
- Fix DPCD register write order to match 128b/132b requirement (Arun)
- Increase AUX timeout for Type-C (Suraj)
- Communicate display power demands to pcode (Stan)
- Fix potential division by zero in DSC compute config (Nikita Zhandarovich)
- Fix fast wake AUX sync length (Jouni)
- Fix potential oops on intel_get_crtc_new_encoder() (Ville)

Merges:
- drm-next backmerges (Rodrigo, Jani)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87zg5eat32.fsf@intel.com

Merge tag 'drm-misc-next-2023-06-07' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for v6.5:

UAPI Changes:

Cross-subsystem Changes:

Core Changes:

Driver Changes:

 * bridge
   * imx: Fix module linking
   * tc358762: Support reset GPIO

 * meson
   * Add support for MIPI DSI displays; plus fixes and DT bindings

 * panel
   * Add Support for Rocktech RK043FN48H; plus DT bindings
   * Add support for Starry himax83102-j02; plus DT bindings
   * Add support for Starry ili9882t; plus DT bindings

 * virtio
   * Support sync-object UAPI

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230607085644.GA12673@linux-uq9g

Merge tag 'drm-habanalabs-next-2023-06-08' of https://git./linux/kernel/git/ogabbay/linux into drm-next

This tag contains additional habanalabs driver changes for v6.5:

- uAPI changes:
  - Return 0 when user queries if there was a h/w or f/w error and no such error happened.
    Previously we returned an error in such case.

- New features and improvements:
  - Add pci health check when we lose connection with the firmware. This can be used to
    distinguish between pci link down and firmware getting stuck.
  - Add more info to the error print when TPC interrupt occur.
  - Reduce amount of code under mutex in the command submission of signal event.

- Firmware related fixes:
  - Fixes to the handshake protocol during f/w initialization.
  - Display information that the f/w sends us when encountering a DMA error.
  - Do soft-reset using a message sent to firmware instead of writing to MMIO.
  - Prepare generic code to extract f/w version numbers.

- Bug fixes and code cleanups. Notable fixes are:
  - Unsecure certain TPC registers that the user should access.
  - Fix handling of QMAN errors
  - Fix memory leak when recording errors (to later pass them to the user)
  - Multiple fixes to razwi interrupt handling code

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Oded Gabbay <ogabbay@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230608103043.GA2699019@ogabbay-vm-u20.habana-labs.com

Merge tag 'drm-next-20230529' of git://git./linux/kernel/git/pinchartl/linux into drm-next

Renesas DRM/KMS drivers:

- Group drivers in renesas subdirectory to prepare for new platform
- Drop deprecated R-Car H3 ES1.x support

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230602111504.GA24855@pendragon.ideasonboard.com

accel/habanalabs: refactor error info reset

Moved error info reset code to single function for future use from
other places in the driver.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add event queue extra validation

In order to increase reliability of the event queue interface,
we apply to Gaudi2 the same mechanism we have in Gaudi1.
The extra validation is basically checking that the received
event index matches the expected index.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: unsecure TSB_CFG_MTRR regs

In order to utilize Engine Barrier padding, user must have access to
this register set.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: move ioctl error print to debug level

We don't want to allow users to spam the kernel log and sending
ioctls with bad opcodes is a sure way to do it.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: fix bug of not fetching addr_dec info

addr_dec info should always be fetched, regardless of cause value.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add description to several info ioctls

Several info ioctls may return success although no data retrieved.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: remove sim code

There were a few places where simulator only code got into the upstream.
Remove those places that can confuse other developers.

Fixes: 2a0a839b6a28 ("habanalabs: extend fatal messages to contain PCI info")
Cc: Moti Haimovski <mhaimovski@habana.ai>
Cc: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: mask part of hmmu page fault captured address

When receiving page fault from hmmu, the captured address is scrambled
both by HW and by driver. The driver part is unscrambled but the HW
part isn't getting unscrambled.
To avoid declaring wrong address, the HW scrambled part will be
masked.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: update state when loading boot fit

Any FW component we load must be followed by a corresponding state
update. However, it seems that so far we skipped doing so for the
bootfit case, so fix that.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: print qman data on error only for lower qman

By default, the upper QMANs are not used, and instead engines ARCs
access the lower QMANs directly.
Errors for upper QMANs are therefore not expected, and the debug print
of the PQ entries is not needed.

Modify the QMAN debug data print on errors to include only information
for the lower QMAN.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: use lower QM in QM errors handling

The QMAN GLBL_ERR_STS_4 register has indications for errors also in the
lower CQ and the ARC CQ, and not just for errors in the lower CP.
Modify the relevant define/struct and the related print to use "lower
QM" instead of "lower CP".

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: use binning info when handling razwi

When receiving sei interrupt from tpc or decoder, we need to check
the binning mask because if the engine is binned, the razwi info
won't be in the router of the binned engine, instead will be in the
router of the substitute engine.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: remove support for mmu disable

As mmu disable mode is only used for bring-up stages, let's remove this
option and all code related to it.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: upon DMA errors, use FW-extracted error cause

Initially, the driver used to read the error cause data directly from
the ASIC. However, the FW now clears it before the driver could read
it. Therefore we should use the error cause data that is extracted by
the FW.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: print max timeout value on CS stuck

If a workload got stuck, we print an error to the kernel log about it.
Add to that print the configured max timeout value, as that value is
not fixed between ASICs and in addition it can be configured using
a kernel module parameter.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: align to latest firmware specs

Update the firmware common interface files with the latest version.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: fix mem leak in capture user mappings

This commit fixes a memory leak caused when clearing the user_mappings
info when a new context is opened immediately after user_mapping is
captured and a hard reset is performed.

Signed-off-by: Moti Haimovski <mhaimovski@habana.ai>
Reviewed-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: set unused bit as reserved

Get latest f/w gaudi2 interface file which marks unused
bist_need_iatu_config bit in cold_rst_data structure as reserved bit.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: rename security functions related arguments

Make the argument names specify the registers array represent
registers that should be unsecured so the user can access them.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix gaudi2_get_tpc_idle_status() return

The gaudi2_get_tpc_idle_status() function returned the incorrect variable
so it always returned true.

Fixes: d85f0531b928 ("accel/habanalabs: break is_idle function into per-engine sub-routines")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: Fix some kernel-doc comments

Make the description of @regs_range_array and @regs_range_array_size
to @user_regs_range_array and @user_regs_range_array_size  to silence
the warnings:

drivers/accel/habanalabs/common/security.c:506: warning: Function parameter or member 'user_regs_range_array' not described in 'hl_init_pb_ranges_single_dcore'
drivers/accel/habanalabs/common/security.c:506: warning: Function parameter or member 'user_regs_range_array_size' not described in 'hl_init_pb_ranges_single_dcore'
drivers/accel/habanalabs/common/security.c:506: warning: Excess function parameter 'regs_range_array' description in 'hl_init_pb_ranges_single_dcore'
drivers/accel/habanalabs/common/security.c:506: warning: Excess function parameter 'regs_range_array_size' description in 'hl_init_pb_ranges_single_dcore'

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Closes: https://bugzilla.openanolis.cn/show_bug.cgi?id=4940
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: always fetch pci addr_dec error info

Due to missing indication of address decode source (LBW/HBW bus),
we should always try and fetch extended information.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix a static warning - 'dubious: x & !y'

Use a straight forward approach to get a conditional result.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: poll for device status update following WFE cmd

Currently, we rely on COMMS protocol's ack to verify that WFE command
has been acknowledged by the FW. However, this does not guarantee that
the device status has been updated.
Although unlikely, this could trigger a race since the driver expects
the device to be halted at that stage, but it might not be.
Therefore, we increase WFE's robustness by polling on the status
register that will be updated once the device is actually halted.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: expose debugfs files later

Currently the debugfs root folder and files for a device are created at
an early step, before the device initialization and before the char
device and sysfs files are exposed to user.
As there is no real reason not to do it together with the device
creation, postpone it to be done right afterwards.

The initialization of the debugfs entry structure is left in its
current position because it is used before creating the files.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add pci health check during heartbeat

Currently upon a heartbeat failure, we don't know if the failure
is due to firmware hang or due to a bad PCI link. Hence, we
are reading a PCI config space register with a known value (vendor ID)
so we will know which of the two possibilities caused the heartbeat
failure.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add missing tpc interrupt info

For some reason the last possible tpc interrupt cause in
gaudi2_tpc_interrupts_cause is missing from the code.

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: refactor abort of completions and waits

Aborting CS completions should be in command_submission.c but aborting
waiting for user interrupts should be in device.c.

This separation is also for adding more abort operations in the future.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: minimize encapsulation signal mutex lock time

Sync Stream Encapsulated Signal Handlers can be managed from different
contexts, and as such they are protected via a spin_lock.
However, spin_lock was unnecessarily protecting a larger code section
than really needed, covering a sleepable code section as well.
Since spin_lock disables preemption, it could lead to sleeping in
atomic context.

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

drm/i915: re-enable -Wunused-but-set-variable

W=1 enables -Wunused-but-set-variable. We disabled it locally in i915
Makefile as we were hitting a bunch of warnings. See commit 6a05d2900464
("drm/i915: Disable unused-but-set compiler warning").

With the issues fixed or annotated with __maybe_unused, re-enable the
warning not only in W=1 but also locally as part of i915 build.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/28ca3f95fe77ceb8aa35b87fca73f7afbc89859a.1685119007.git.jani.nikula@intel.com

drm/i915/gvt: remove unused variable gma_bottom in command parser

Remove unused variable gma_bottom in scan_workload() and scan_wa_ctx().
commit be1da7070aea ("drm/i915/gvt: vGPU command scanner") introduces
gma_bottom in several functions to calculate the size of the command
buffer. However, some of them are set but actually unused.

When compiling the code with ccflags -Wunused-but-set-variable, gcc
throws warnings.

Remove unused variables to avoid the gcc warnings. Tested via compiling
the code with ccflags -Wunused-but-set-variable.

Fixes: be1da7070aea ("drm/i915/gvt: vGPU command scanner")
Suggested-by: Jani Nikula <jani.nikula@intel.com>
Cc: Zhenyu Wang <zhenyuw@linux.intel.com>
Cc: intel-gvt-dev@lists.freedesktop.org
Signed-off-by: Zhi Wang <zhi.a.wang@intel.com>
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230531020411.18987-1-zhi.a.wang@intel.com

drm/i915/gsc: Fix error code in intel_gsc_uc_heci_cmd_submit_nonpriv()

This should return negative -EAGAIN instead of positive EAGAIN.

Fixes: e5e1e6d28ebc ("drm/i915/pxp: Add MTL helpers to submit Heci-Cmd-Packet to GSC")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Alan Previn <alan.previn.teres.alexis@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZH7sr+Vs4zOQoouU@moroto

drm/i915/display: Include of display limits doesn't need 'display/'

Drop the unnecessary directory prefix.  This also makes
intel_display_device.h easier to use from the Xe driver.

Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230607150946.1996087-1-matthew.d.roper@intel.com

i915/perf: Do not add ggtt offset to hw_tail

ggtt offset for hw_tail is not required for the calculations, so drop
it.

Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230605193923.1836048-3-umesh.nerlige.ramappa@intel.com

i915/perf: Drop the aging_tail logic in perf OA

On DG2, capturing OA reports while running heavy render workloads
sometimes results in invalid OA reports where 64-byte chunks inside
reports have stale values. Under memory pressure, high OA sampling rates
(13.3 us) and heavy render workload, occasionally, the OA HW TAIL
pointer does not progress as fast as the sampling rate. When these
glitches occur, the TAIL pointer takes approx. 200us to progress.  While
this is expected behavior from the HW perspective, invalid reports are
not expected.

In oa_buffer_check_unlocked(), when we execute the if condition, we are
updating the oa_buffer.tail to the aging tail and then setting pollin
based on this tail value, however, we do not have a chance to rewind and
validate the reports prior to setting pollin. The validation happens
in a subsequent call to oa_buffer_check_unlocked(). If a read occurs
before this validation, then we end up reading reports up until this
oa_buffer.tail value which includes invalid reports. Though found on
DG2, this affects all platforms.

The aging tail logic is no longer necessary since we are explicitly
checking for landed reports.

Start by dropping the aging tail logic.

v2:
- Drop extra blank line
- Add reason to drop aging logic (Ashutosh)
- Add bug links (Ashutosh)
- rename aged_tail to read_tail
- Squash patches 3 and 1

v3: (Ashutosh)
- Remove extra spaces
- Remove gtt_offset from the pollin calculation
- s/Bug:/Link/ in commit message (checkpatch)

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/7484
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/7757
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230605193923.1836048-2-umesh.nerlige.ramappa@intel.com

drm/i915/mtl: Add support for PM DEMAND

MTL introduces a new way to instruct the PUnit with
power and bandwidth requirements of DE. Add the functionality
to program the registers and handle waits using interrupts.
The current wait time for timeouts is programmed for 10 msecs to
factor in the worst case scenarios. Changes made to use REG_BIT
for a register that we touched(GEN8_DE_MISC_IER _MMIO).

Wa_14016740474 is added which applies to Xe_LPD+ display

v2: checkpatch warning fixes, simplify program pmdemand part

v3: update to dbufs and pipes values to pmdemand register(stan)
    Removed the macro usage in update_pmdemand_values()

v4: move the pmdemand_pre_plane_update before cdclk update
    pmdemand_needs_update included cdclk params comparisons
    pmdemand_state NULL check (Gustavo)
    pmdemand.o in sorted order in the makefile (Jani)
    update pmdemand misc irq handler loop (Gustavo)
    active phys bitmask and programming correction (Gustavo)

v5: simplify pmdemand_state structure
    simplify methods to find active phys and max port clock
    Timeout in case of previou pmdemand task pending (Gustavo)

v6: rebasing
    updates to max_ddiclk calculations (Gustavo)
    updates to active_phys count method (Gustavo)

v7: use two separate loop to iterate throug old and new
    crtc states to calculate the active phys (Gustavo)

v8: use uniform function names (Gustavo)

v9: For phys change iterate through connectors (Imre)
    Look for change in phys for pmdemand update (Gustavo, Imre)
    Some more stlying changes (Imre)
    Update pmdemand state during HW readout/sanitize (Imre)

v10: Fix CI checkpatch warnings

v11: use correct pmdemand object pointer during hw readout,
     simplify the check for phys need update (Gustavo)

v12: Handle possible non serialize cases (Imre)
     Initialise also pmdemand params HW readout (Imre)
     Update active phys mask during sanitize calls (Imre)
     Check TC/encoder changes to limit connector update (Imre)

v13: Check display version before accessing pmdemand functions

v14: Move is_serialized to intel_global_state.c
     simplify update params and other stlying issues (Imre)

Bspec: 66451, 64636, 64602, 64603
Cc: Matt Atwood <matthew.s.atwood@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com> #v4
Acked-by: Gustavo Sousa <gustavo.sousa@intel.com> #v11
Reviewed-by: Imre Deak <imre.deak@intel.com>
[RK: Fixed minor typo in one of the comments. s/qclck_gc/qclk_gv/]
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606201032.347449-1-vinod.govindapillai@intel.com

drm/i915/mtl: find the best QGV point for the SAGV configuration

From MTL onwards, we need to find the best QGV point based on
the required data rate and pass the peak BW of that point to
the punit to lock the corresponding QGV point.

v1: Fix for warning from kernel test robot

v2: No need to serialize for the peakbw change as pmdemand code
    will do that (Imre)

Bspec: 64636

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/r/202305280253.Ab8bRV2w-lkp@intel.com/
Reported-by: Dan Carpenter <error27@gmail.com>
Closes: https://lore.kernel.org/r/202305280253.Ab8bRV2w-lkp@intel.com/
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606093509.221709-7-vinod.govindapillai@intel.com

drm/i915: modify max_bw to return index to intel_bw_info

MTL uses the peak BW of a QGV point to lock the required QGV
point instead of the QGV index. Instead of passing the deratedbw
of the selected bw_info, return the index to the selected
bw_info so that either deratedbw or peakbw can be used based on
the platform.

v2: use idx to store index returned by max_bw_index functions

v3: return UINT_MAX in icl_max_bw_index in case no match found

v3: check idx >= ARRAY_SIZE

Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606093509.221709-6-vinod.govindapillai@intel.com

drm/i915: extract intel_bw_check_qgv_points()

Extract intel_bw_check_qgv_points() from intel_bw_atomic_check
to facilitate future platform variations in handling SAGV
configurations.

Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606093509.221709-5-vinod.govindapillai@intel.com

drm/i915: store the peak bw per QGV point

In MTL onwards, pcode locks the GV point based on the peak BW
of a QGV point. So store the peak BW of all the QGV points.

v2: use DIV_ROUND_CLOSEST() for the peakBW calculation

Bspec: 64636

Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606093509.221709-4-vinod.govindapillai@intel.com

drm/i915: update the QGV point frequency calculations

From MTL onwwards, pcode locks the QGV point based on peak BW of
the intended QGV point passed by the driver. So the peak BW
calculation must match the value expected by the pcode. Update
the calculations as per the Bspec.

v2: use DIV_ROUND_* macro for the calculations (Ville)

v3: Use only DIV_ROUN_CLOSEST and remove divisor / 2 again

Bspec: 64636

Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606093509.221709-3-vinod.govindapillai@intel.com

drm/i915: fix the derating percentage for MTL

Follow the values from bspec for the percentage overhead for
efficiency in MTL BW calculations.

Bspec: 64631

Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230606093509.221709-2-vinod.govindapillai@intel.com