A DMC bug on GEN9 big core machines fails to restore the driver's
request bits for the PW1 and MISC_IO power wells after a DC5/6
entry->exit sequence. As a consequence the driver's subsequent check for
the enabled status of these power wells will fail, as the check
considers the power wells being enabled only if both the status and
request bits are set. To work around this borrow the request bits from
BIOS's own request register in which DMC forces on the request bits when
exiting from DC5/6.
This fixes a problem reported by Ramalingam, where HDCP init failed,
since PW1 reported itself as being disabled, while in reality it was
enabled.
Reported-by: Ramalingam C <ramalingam.c@intel.com>
Cc: Ramalingam C <ramalingam.c@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181109145822.15446-1-imre.deak@intel.com
This got duplicated on introducing icl workarounds.
Fix by using the older definition and moving the wa bit
definition there. No functional changes.
v3: avoid fixes tag, whitespace (Chris)
References: 908ae05173 ("drm/i915/icl: WaDisCtxReload")
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20181109145333.10570-1-mika.kuoppala@linux.intel.com
Register DBUF_CTL_S2 is read and it's value is not used. As
there is no explanation why we should prime the hardware with
read, remove it as spurious.
Fixes: aa9664ffe8 ("drm/i915/icl: Enable 2nd DBuf slice only when needed")
Cc: Mahesh Kumar <mahesh1.kumar@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181109140924.2663-1-mika.kuoppala@linux.intel.com
I think I'm probably the one who argued in favor of having separate
implementations for both PCHs, but the calculations are actually the
same, the clocks are the same and the only difference is that on ICP
we write the numerator to the register.
I have previously suggested to kill cnp_rawclk() and keep the
icp_rawclk() style, but Ville gave some good arguments that what's in
this patch may be the better choice.
v2: Switch numerator to 1 from 1000 and adjust calculations
accordingly (Ville).
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181112232313.26373-3-paulo.r.zanoni@intel.com
Although CNP names this field "Counter Fraction", what we write to the
register is really the denominator for the fractional part of the
divider, not the fractional part (and the field description even says
that). The ICP spec renamed the field to "Counter Fraction
Denominator", which makes a lot more sense. Use the more complete ICL
naming because we will merge the CNP and ICP functions into a single
one, which will introduce the concept of the numerator. That will make
a lot more sense when you read the "num/frac = den" calculation.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181112232313.26373-2-paulo.r.zanoni@intel.com
BSpec was updated and now there's no more "subtract 1" to the
Microsecond Counter Divider field.
It seems this should help fixing some GMBUS issues. I'm not aware of
any specific open bug that could be solved by this patch.
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181112232313.26373-1-paulo.r.zanoni@intel.com
VBT appears to have two (or possibly three) ways to indicate the panel
rotation. The first is in the MIPI config block, but that apparenly
usually (maybe always?) indicates 0 degrees despite the actual panel
orientation. The second way to indicate this is in the general features
block, which can just indicate whether 180 degress rotation is used.
The third might be a separate rotation data block, but that is not
at all documented so who knows what it may contain.
Let's try the first two. We first try the DSI specicic VBT
information, and it it doesn't look trustworthy (ie. indicates
0 degrees) we fall back to the 180 degree thing. Just to avoid too
many changes in one go we shall also keep the hardware readout path
for now.
If this works for more than just my VLV FFRD the question becomes
how many of the panel orientation quirks are now redundant?
v2: Move the code into intel_dsi.c (Jani)
Cc: Hans de Goede <hdegoede@redhat.com>
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/20181022142015.4026-1-ville.syrjala@linux.intel.com
Tested-by: Hans de Goede <hdegoede@redhat.com>
Let's make sure the DSI port is actually on before we go
poking at the plane register to determine which way
it's rotated. Otherwise we could be looking at a plane
that is feeding a HDMI port for instance.
And in order to read the plane register we need the power
well to be on. Make sure that is indeed the case. We'll
also make sure the plane is actually enabled before we
trust the rotation bit to tell us the truth.
v2: s/intel_dsi/vlv_dsi/
Cc: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181022141953.3889-1-ville.syrjala@linux.intel.com
Tested-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> #irc
No point in cluttering the common codepaths with the
skip_intermediate_wm handling. Just move it into
ilk_compute_intermediate_wm() as those are the only
platforms using this.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108151013.24064-1-ville.syrjala@linux.intel.com
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> #irc
To get the initial phase correct we need to account for the scale
factor as well. I forgot this initially and was mostly looking at
heavily upscaled content where the minor difference between -0.5
and the proper initial phase was not readily apparent.
And let's toss in a comment that tries to explain the formula
a little bit.
v2: The initial phase upper limit is 1.5, not 24.0!
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Fixes: 0a59952b24 ("drm/i915: Configure SKL+ scaler initial phase correctly")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181029181820.21956-1-ville.syrjala@linux.intel.com
Tested-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Tested-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> #irc
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> #irc
Reduce the clutter in the sprite update functions by writing
both TILEOFF and LINOFF registers unconditionally. We already
did this for primary planes so might as well do it for the
sprites too.
There is no harm in writing both registers. Which one gets
used depends on the tilimg mode selected in the plane control
registers.
It might even make sense to clear the register that won't
get used. That could make register dumps a little easier to
parse. But I'm not sure it's worth the extra hassle.
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108150955.23948-1-ville.syrjala@linux.intel.com
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> #irc
We no longer change LSPCON into PCON mode if it boots up in
LS mode. This was broken by some code shuffling in
commit 96e35598ce ("drm/i915: Check LSPCON vendor OUI").
I actually can't see a reason why that code shuffling had
to be done. The commit msg notes it but doesn't justify it
in any way. But I guess we'll keep the code in its current
place anyway and just make the "switch to PCON mode" part
effective once again.
Cc: Shashank Sharma <shashank.sharma@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Tested-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Fixes: 96e35598ce ("drm/i915: Check LSPCON vendor OUI")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107171821.27862-1-ville.syrjala@linux.intel.com
Reviewed-by: Shashank Sharma <shashank.sharma@intel.com>
Pass on the errno all the way from connected_sink_max_bpp(),
and make the base_bpp handling in intel_modeset_pipe_config()
a bit less ugly. We'll also rename connected_sink_max_bpp()
to not give the impression that it return the bpp value,
and we'll pimp up the debug message within to include the
connector name/id.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107213522.17590-3-ville.syrjala@linux.intel.com
Reviewed-by: Imre Deak <imre.deak@intel.com>
ironlake_check_fdi_lanes() may try to grab some extra crtc locks.
If that fails we need to propagate the -EDEADLK all the way up,
and we shouldn't dump out the crtc state or other debug messages
either since it wasn't the crtc state that caused the failure.
Just hit this on my IVB:
[drm:intel_atomic_check [i915]] checking fdi config on pipe C, lanes 3
[drm:intel_atomic_check [i915]] only 2 lanes on pipe C: required 3 lanes
[drm:intel_atomic_check [i915]] fdi link bw constraint, reducing pipe bpp to 18
[drm:intel_atomic_check [i915]] checking fdi config on pipe C, lanes 2
[drm:intel_atomic_check [i915]] CRTC bw constrained, retrying
[drm:intel_dp_compute_config [i915]] DP link computation with max lane count 4 max rate 270000 max bpp 18 pixel clock 185580KHz
[drm:intel_dp_compute_config [i915]] DP lane count 4 clock 162000 bpp 18
[drm:intel_dp_compute_config [i915]] DP link rate required 417555 available 648000
[drm:intel_atomic_check [i915]] checking fdi config on pipe C, lanes 2
WARNING: CPU: 4 PID: 25115 at ../drivers/gpu/drm/drm_modeset_lock.c:241 drm_modeset_lock+0xbc/0xd0 [drm]
...
WARNING: CPU: 4 PID: 25115 at ../drivers/gpu/drm/drm_modeset_lock.c:223 drm_modeset_drop_locks+0x4a/0x50 [drm]
The warnings are from 'WARN_ON(ctx->contended)'.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107213522.17590-1-ville.syrjala@linux.intel.com
Reviewed-by: Imre Deak <imre.deak@intel.com>
We dump the info as an array of u8, so we want to know the length
in number of bytes. Current code is still safe because the
variable we use BITS_PER_TYPE on is a u8.
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181109004013.34394-1-daniele.ceraolospurio@intel.com
We have a subslice mask per slice, not per subslice.
MAX_SUBSLICES > MAX_SLICES, so the wrong size didn't cause any issue
apart from using extra memory.
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106182918.5748-1-daniele.ceraolospurio@intel.com
In the past we had hooks to configure HW for VLV/CHV too, in the drop
of VLV/CHV support the intel_psr_disable_source() code was not moved
to the caller, so doing it here.
Suggested-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106190843.18009-4-jose.souza@intel.com
All other interruptions gen11 interruptions are reset in
gen11_irq_reset() also it is done for other gens that supports PSR.
Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106190843.18009-3-jose.souza@intel.com
It should always wait for idle state when disabling PSR because PSR
could be inactive due a call to intel_psr_exit() and while PSR is
still being disabled asynchronously userspace could change the
modeset causing a call to psr_disable() that will not wait for PSR
idle and then PSR will be enabled again while PSR is still not idle.
v2: rebased on top of the patch reusing psr_exit()
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106190843.18009-2-jose.souza@intel.com
Both functions have the same code to disable PSR, so let's reuse that
code instead of duplicate.
Suggested-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106190843.18009-1-jose.souza@intel.com
Make skl_ddb_allocation_overlaps() useful for other callers
besides skl_update_crtcs(). We'll need it to do plane updates
as well.
And while we're here we can reduce the stack utilization a
bit by noting that each struct skl_ddb_entry is 4 bytes whereas
a pointer to one is 8 bytes (on 64bit). So we'll switch to an
array of structs from the array of pointers we used before.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181101150605.18235-9-ville.syrjala@linux.intel.com
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Due to the constant alpha we're going to have to program two of
the the tree keying registers anyway, so might as well always
program all three.
And parametrize the plane constant alpha define while at it.
v2: Rebase due to input CSC
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107184138.31359-1-ville.syrjala@linux.intel.com
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
MST is only supported in DDI ports that have this hook, so the null
check can be dropped.
Suggested-by: Imre Deak <imre.deak@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107235449.32264-3-jose.souza@intel.com
MST ports did not had the post_pll_disable() hook causing the
references get in pre_pll_enable() never being released causing
DDI and AUX CH being enabled all the times.
v2: renamed intel_mst_post_pll_disable_dp() parameters
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Manasi Navare <manasi.d.navare@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107235449.32264-2-jose.souza@intel.com
intel_dp_detect() caches the aux_domain in the beginning of the
function as it is used twice, so lets also use it as the aux_domain
don't change in runtime.
v3: returning intel_dp_retrain_link() error insted of
connector_status_disconnected
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107235449.32264-1-jose.souza@intel.com
IBX has a documented workaround which states that when we disable the
port we must change its transcoder select to A, otherwise it will
prevent the other port (DP vs. HDMI/SDVO) from using transcoder A.
We implement the workaround during encoder disable, but looks like
some BIOSen leave transcoder B selected even when the port wasn't
actually enabled by the BIOS. That will trip up our asserts
that attempt to make sure we never forget this w/a.
Sanitize the transcoder select to A for all disabled PCH
DP/HDMI/SDVO ports. We assume that the port was never enabled
by the BIOS on transcoder B, because if it had we'd actually have
to toggle the port on and back off to properly switch it back to
transcoder A. That would cause some display flicker if transcoder A
is already enabled on some other port, so it's better not to do it
unless absolutely necessary. Since we have no indication that the
transcoder select is misbehaving on the affected machines we can
assume the port was never actually enabled by the BIOS.
This cures warning like this during driver load:
IBX PCH DP C still using transcoder B
WARNING: CPU: 2 PID: 172 at drivers/gpu/drm/i915/intel_display.c:1279 assert_pch_dp_disabled+0x9e/0xb0 [i915]
v2: Add comments to remind the reader that SDVOB==HDMIB (Chris)
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108143635.9556-1-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
In my haste to remove irq_port[] I accidentally changed the
way we deal with hpd pins that are shared by multiple encoders
(DP and HDMI for pre-DDI platforms). Previously we would only
handle such pins via ->hpd_pulse(), but now we queue up the
hotplug work for the HDMI encoder directly. Worse yet, we now
count each hpd twice and this increment the hpd storm count
twice as fast. This can lead to spurious storms being detected.
Go back to the old way of doing things, ie. delegate to
->hpd_pulse() for any pin which has an encoder with that hook
implemented. I don't really like the idea of adding irq_port[]
back so let's loop through the encoders first to check if we
have an encoder with ->hpd_pulse() for the pin, and then go
through all the pins and decided on the correct course of action
based on the earlier findings.
I have occasionally toyed with the idea of unifying the pre-DDI
HDMI and DP encoders into a single encoder as well. Besides the
hotplug processing it would have the other benefit of preventing
userspace from trying to enable both encoders at the same time.
That is simply illegal as they share the same clock/data pins.
We have some testcases that will attempt that and thus fail on
many older machines. But for now let's stick to fixing just the
hotplug code.
Cc: stable@vger.kernel.org # 4.19+
Cc: Lyude Paul <lyude@redhat.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Fixes: b6ca3eee18 ("drm/i915: Nuke dev_priv->irq_port[]")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108200424.28371-1-ville.syrjala@linux.intel.com
Reviewed-by: Lyude Paul <lyude@redhat.com>
We shouldn't consider an encoder inactive if it doesn't have a CRTC
linked, but has virtual MST encoders with a crtc linked. Fix this.
Also we should not sanitize the mapping for MST encoders, as it's always
their primary encoder (which could be even in SST mode) whose active
state determines if we need the clock being enabled for the
corresponding physical port. Fix this too.
This fixes at least an existing breakage where we incorrectly disabled
the clock for an active DP encoder when sanitizing its MST virtual
encoders. Not sure if there are BIOSes that enable an output in MST
mode, but our HW readout is mostly missing for it anyway, so just warn
for that case.
Fixes: 70332ac539 ("drm/i915/icl+: Sanitize port to PLL mapping")
Cc: Antonio Argenziano <antonio.argenziano@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reported-by: Antonio Argenziano <antonio.argenziano@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Tested-by: Clint Taylor <Clinton.A.Taylor@intel.com>
Reviewed-by: Clint Taylor <Clinton.A.Taylor@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107200836.10191-2-imre.deak@intel.com
Check for reserved register field values and conflicting
transcoder->port mappings (both MST and non-MST mappings or multiple SST
mappings).
This is also needed for the next patch to determine if a port is in MST
mode during sanitization after HW readout.
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Antonio Argenziano <antonio.argenziano@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Clint Taylor <Clinton.A.Taylor@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181107200836.10191-1-imre.deak@intel.com
While our little rcu worker might be able to be replaced now by the
dedicated rcu_work, in the meantime we should mark up the rcu_head for
correct debugobjects tracking.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108092101.27598-2-chris@chris-wilson.co.uk
All other overlay functions(almost all other functions in i915)
follow intel_overlay_verb, so renaming overlay ones that do not match
that.
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108001647.11276-4-jose.souza@intel.com
IPC is a display feature, so i915_load_modeset_init() is the right
place to initialize it.
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108001647.11276-3-jose.souza@intel.com
Although FBC helps save power it do not belongs to power management
also the cleanup was placed in i915_driver_unload() also not a good
place. intel_modeset_init()/intel_modeset_cleanup() are better places
also this will help make easy disable features that depends in
display being enabled in driver.
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108001647.11276-2-jose.souza@intel.com
i915_load_modeset_init() is a more suitable place than
i915_driver_load() as vblank is part of modeset.
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108001647.11276-1-jose.souza@intel.com
This reduces the size of struct skl_wm_level from 6 to 4, which
reduces the size of struct skl_plane_wm from 104 to 70, which reduces
the size of struct skl_pipe_wm from 524 to 356. A reduction of 168
padding bytes per pipe. This will increase even more the next time we
bump I915_MAX_PLANES.
v2: Paste the pahole output provided by Lucas:
$ pahole -s -C skl_wm_level drivers/gpu/drm/i915/i915.o
struct skl_wm_level {
bool plane_en; /* 0 1 */
/* XXX 1 byte hole, try to pack */
uint16_t plane_res_b; /* 2 2 */
uint8_t plane_res_l; /* 4 1 */
/* size: 6, cachelines: 1, members: 3 */
/* sum members: 4, holes: 1, sum holes: 1 */
/* padding: 1 */
/* last cacheline: 6 bytes */
};
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181016220133.26991-3-paulo.r.zanoni@intel.com
We have multiple instances of VCS but only remember to invalidate the
BSD caches on the first, ignoring the stale caches of any other engine.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181108140039.12254-1-chris@chris-wilson.co.uk
On ICL DMC/PCODE retains the HW context only for port A across DC
transitions, for the other port B combo PHY, it doesn't. So we need to
do this manually after exiting from DC6. Do the reinit even after
exiting from DC5, it won't hurt since we only reinit the PHY in case
it's needed (in case it was disabled to begin with).
As can be guessed from the bugzilla report leaving the PHY uninited will
lead to a later timeout during the port B specific AUX and DDI_IO power
well enabling.
v2:
- Apply the fix on all GEN>=11 platforms. (Rodrigo)
Bspec: 21257
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108070
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106160621.23057-6-imre.deak@intel.com
Bspec says we should skip the initialization of combo PHYs that are
already initialized. We'll need to reinit the PHYs more frequently
when exiting from DC6 (after the next patch), so let's make sure the
uninit sequence complies with the spec. For safety skip the init only if
all the PHY register fields have their expected values.
v2:
- Print 'Port X' as we do elsewhere instead of 'Port-X'. (Jose)
Bspec: 21257
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106160621.23057-5-imre.deak@intel.com
Verify on CNL, ICL that the combo PHY HW state stayed intact after PHY
initialization.
v2:
- Print 'Port X' as we do elsewhere instead of 'Port-X'. (Jose)
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106160621.23057-4-imre.deak@intel.com
Similarly to the GEN9_LP DPIO PHY code keep the CNL+ combo PHY code in a
separate file.
No functional change.
v2:
- Use SPDX license tag instead of boilerplate. (Rodrigo)
v3:
- Use MIT instead of GPL-2.0 license. (Ville)
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106160621.23057-3-imre.deak@intel.com
BSpec says to clear the comp init HW flag too during combo PHY uninit,
so do that. The lack of this could badly interact with the PHY reinit
after a DC6/9 transition at least, where (after a follow-up patch fixing
the init code) we'd skip the initialization incorrectly due to this flag
being set.
BSpec: 21257
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106160621.23057-2-imre.deak@intel.com
To enable DC5/6 power well 2 has to be disabled as for previous
platforms, so fix things up.
Bspec: 4234
Fixes: 67ca07e7ac ("drm/i915/icl: Add power well support")
Cc: Animesh Manna <animesh.manna@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181102182200.17219-1-imre.deak@intel.com
Unfortunately, it seems that the HPD IRQ storm problem from the early
days of Intel GPUs was never entirely solved, only mostly. Within the
last couple of days, I got a bug report from one of our customers who
had been having issues with their machine suddenly booting up very
slowly after having updated. The amount of time it took to boot went
from around 30 seconds, to over 6 minutes consistently.
After some investigation, I discovered that i915 was reporting massive
amounts of short HPD IRQ spam on this system from the DisplayPort port,
despite there not being anything actually connected. The symptoms would
start with one "long" HPD IRQ being detected at boot:
[ 1.891398] [drm:intel_get_hpd_pins [i915]] hotplug event received, stat 0x00440000, dig 0x00440000, pins 0x000000a0
[ 1.891436] [drm:intel_hpd_irq_handler [i915]] digital hpd port B - long
[ 1.891472] [drm:intel_hpd_irq_handler [i915]] Received HPD interrupt on PIN 5 - cnt: 0
[ 1.891508] [drm:intel_hpd_irq_handler [i915]] digital hpd port D - long
[ 1.891544] [drm:intel_hpd_irq_handler [i915]] Received HPD interrupt on PIN 7 - cnt: 0
[ 1.891592] [drm:intel_dp_hpd_pulse [i915]] got hpd irq on port B - long
[ 1.891628] [drm:intel_dp_hpd_pulse [i915]] got hpd irq on port D - long
…
followed by constant short IRQs afterwards:
[ 1.895091] [drm:intel_encoder_hotplug [i915]] [CONNECTOR:66:DP-1] status updated from unknown to disconnected
[ 1.895129] [drm:i915_hotplug_work_func [i915]] Connector DP-3 (pin 7) received hotplug event.
[ 1.895165] [drm:intel_dp_detect [i915]] [CONNECTOR:72:DP-3]
[ 1.895275] [drm:intel_get_hpd_pins [i915]] hotplug event received, stat 0x00200000, dig 0x00200000, pins 0x00000080
[ 1.895312] [drm:intel_hpd_irq_handler [i915]] digital hpd port D - short
[ 1.895762] [drm:intel_get_hpd_pins [i915]] hotplug event received, stat 0x00200000, dig 0x00200000, pins 0x00000080
[ 1.895799] [drm:intel_hpd_irq_handler [i915]] digital hpd port D - short
[ 1.896239] [drm:intel_dp_aux_xfer [i915]] dp_aux_ch timeout status 0x71450085
[ 1.896293] [drm:intel_get_hpd_pins [i915]] hotplug event received, stat 0x00200000, dig 0x00200000, pins 0x00000080
[ 1.896330] [drm:intel_hpd_irq_handler [i915]] digital hpd port D - short
[ 1.896781] [drm:intel_get_hpd_pins [i915]] hotplug event received, stat 0x00200000, dig 0x00200000, pins 0x00000080
[ 1.896817] [drm:intel_hpd_irq_handler [i915]] digital hpd port D - short
[ 1.897275] [drm:intel_get_hpd_pins [i915]] hotplug event received, stat 0x00200000, dig 0x00200000, pins 0x00000080
The customer's system in question has a GM45 GPU, which is apparently
well known for hotplugging storms.
So, workaround this impressively broken hardware by changing the default
HPD storm threshold from 5 to 50. Then, make long IRQs count for 10, and
short IRQs count for 1. This makes it so that 5 long IRQs will trigger
an HPD storm, and on systems with short HPD storm detection 50 short
IRQs will trigger an HPD storm. 50 short IRQs amounts to 100ms of
constant pulsing, which seems like a good middleground between being too
sensitive and not being sensitive enough (which would cause visible
stutters in userspace every time a storm occurs).
And just to be extra safe: we don't enable this by default on systems
with MST support. There's too high of a chance of MST support triggering
storm detection, and systems that are new enough to support MST are a
lot less likely to have issues with IRQ storms anyway.
As a note: this patch was tested using a ThinkPad T450s and a Chamelium
to simulate the short IRQ storms.
Changes since v1:
- Don't use two separate thresholds, just make long IRQs count for 10
each and short IRQs count for 1. This simplifies the code a bit
- Ville Syrjälä
Changes since v2:
- Document @long_hpd in intel_hpd_irq_storm_detect, no functional
changes
Changes since v4:
- Remove !! in long_hpd assignment - Ville Syrjälä
- queue_hp = true - Ville Syrjälä
Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106213017.14563-6-lyude@redhat.com
