Now that everything that controls which LRU an obj lives in *except* the
backing pages is protected by the LRU lock, add a special path to unpin
in the job_run() path, where we are assured that we already have backing
pages and will not be racing against eviction (because the GEM object's
dma_resv contains the fence that will be signaled when the submit/job
completes).
Signed-off-by: Rob Clark <robdclark@chromium.org>
Patchwork: https://patchwork.freedesktop.org/patch/527845/
Link: https://lore.kernel.org/r/20230320144356.803762-10-robdclark@gmail.com
Since the LRU lock is already acquired when moving an obj between LRUs,
we can use it to protect pin_count and madv, without any significant
change in locking (ie. it just expands the scope of the lock by a hand-
ful of instructions). This prepares the way to decrement the pin_count
in the job_run() path without needing to hold the obj lock, to avoid a
potential deadlock (or rather stall) caused by the fence-signaling path
(job_run()) blocking on shrinker/reclaim. (Only a stall because the
wait for fence signaling wait_for_idle() is not infinite.)
Signed-off-by: Rob Clark <robdclark@chromium.org>
Patchwork: https://patchwork.freedesktop.org/patch/527843/
Link: https://lore.kernel.org/r/20230320144356.803762-9-robdclark@gmail.com
The flags are only accessed (1) when submit is constructed, before
enqueuing to gpu sched (ie. when still visible to only the task calling
the submit ioctl), (2) here, where we own a reference to the submit and
are serialized on the gpu sched thread, and (3) after the submit is
retired and last reference is dropped, which is serialized on the
submit's reference count. Hence locking is unneeded here.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Patchwork: https://patchwork.freedesktop.org/patch/527830/
Link: https://lore.kernel.org/r/20230320144356.803762-3-robdclark@gmail.com
The DORCR register controls the routing of clocks and data between DU
channels within a group. For groups that contain a single channel,
there's no routing option to control, and some fields of the register
are then reserved. On Gen2 those reserved fields are documented as
required to be set to 0, while on Gen3 and newer the PG1T, DK1S and PG1D
reserved fields must be set to 1.
The DU driver initializes the DORCR register in rcar_du_group_setup(),
where it ignores the PG1T, DK1S and PG1D, and then configures those
fields to the correct value in rcar_du_group_set_routing(). This hasn't
been shown to cause any issue, but prevents certifying that the driver
complies with the documentation in safety-critical use cases.
As there is no reasonable change that the documentation will be updated
to clarify that those reserved fields can be written to 0 temporarily
before starting the hardware, make sure that the registers are always
set to valid values.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The DORCR fields were documented in the R-Car H1 datasheet with 1-based
named, and then got renamed to 0-based in Gen2. The 0-based names are
used for Gen3 and Gen4, making H1 an outlier. Rename the field macros to
make them 0-based, in order to increase readability of the code when
comparing it with the documentation.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
When the input to a DU channel comes from a VSP, the DU doesn't perform
any blending operation. Select XRGB8888 instead of ARGB8888 to ensure
that the corresponding registers don't get written with invalid values.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The ESCR and OTAR registers are not present in all DU channels on Gen3
SoCs. ESCR only exists in channels that can be routed to an LVDS or
DPAD, and OTAR in channels that can be routed to a DPAD. Skip writing
those registers for other channels. This replaces the DU gen check, as
Gen4 doesn't have LVDS or DPAD outputs.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
On R-Car D3 and E3, the LVDS encoder provides the dot (pixel) clock to
the DU, regardless of whether the LVDS output is used or not. When using
the DPAD (RGB) output, the DU driver thus enables and disables the LVDS
PLL manually, while when using the LVDS output, it lets the LVDS bridge
driver handle the PLL configuration internally as part of the atomic
enable and disable operations.
This causes an issue when using the LVDS output. As bridges are disabled
before CRTCs, the current implementation violates the enable/disable
sequences documented in the hardware datasheet, which requires the dot
clock to be enabled before the CRTC is started and disabled after it
gets stopped.
Fix the problem by enabling/disabling the LVDS PLL manually from the DU
regardless of which output is used, and skipping the PLL handling in the
LVDS bridge atomic enable and disable operations.
This is however not enough. Disabling the LVDS encoder while leaving the
PLL on still results in a vertical blanking wait timeout when disabling
the DU. Investigation showed that the culprit is the LVEN bit. For an
unclear reason, clearing the bit when disabling the LVDS encoder blocks
vertical blanking interrupts. We thus have to delay disabling the whole
LVDS encoder, not just disabling the PLL, until the DU is disabled.
We could split the LVDS disable sequence by clearing the LVRES bit in
the LVDS bridge atomic disable handler, and delaying the rest of the
operations, in order to disable the LVDS output at bridge atomic disable
time, before stopping the CRTC. This would make the code more complex,
without a clear benefit, so keep the implementation simple(r).
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen+renesas@ideasonboard.com>
To prepare for a rework of the LVDS disable code, which will need to be
called from rcar_lvds_pclk_disable(), move the LVDS enable code,
currently stored in the __rcar_lvds_atomic_enable() function, to a
separate code section separate from bridge operations. It will be then
extended with the LVDS disable code.
As part of this rework the __rcar_lvds_atomic_enable() function is
renamed to rcar_lvds_enable() to more clearly indicate its purpose.
No functional change intended.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen+renesas@ideasonboard.com>
When disabling the companion bridge in rcar_lvds_atomic_disable(),
there's no need to go through the bridge's operations to call
.atomic_disable(). Call rcar_lvds_atomic_disable() on the companion
directly.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen+renesas@ideasonboard.com>
drm-misc-next for v6.4-rc1:
Core Changes:
- Add unit test for xrgb8888 to mono.
- Assorted small fixes to format helper selftests.
- Assorted documentation updates.
- Drop drm_dev_set_unique.
- Always use shadow buffer in generic fbdev emulation helpers, and
improve error handling.
Driver Changes:
- Assorted small fixes to malidp, hdlcd, gma500, lima, bridge, rockchip.
- Move fbdev in gma500 to use drm_client.
- Convert bridge platform callbacks to void return.
- Drop leftover from vgem to shmem helper conversion.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/5a4c438e-7047-c044-fc77-5a3597000264@linux.intel.com
Now that we may have multiple OA units in a single GT as well as on
separate GTs, create an engine group that maps to a single OA unit.
v2: (Jani)
- Drop warning on ENOMEM
- Reorder patch in the series
v3: (Ashutosh)
- Remove unused members from perf structs
- Update comments
- Update engine_supports_oa check
- Just return 1 in num_perf_groups_per_gt for now
- Set engine->oa_group to NULL to begin with
v4: Use engine_supports_oa() check in oa_init_reg_state (Ashutosh)
v5: Rebase after dropping engine_supports_oa helper
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/20230323225901.3743681-5-umesh.nerlige.ramappa@intel.com
Currently i915_gem_object_is_framebuffer() doesn't treat the
BO containing the framebuffer's DPT as a framebuffer itself.
This means eg. that the shrinker can evict the DPT BO while
leaving the actual FB BO bound, when the DPT is allocated
from regular shmem.
That causes an immediate oops during hibernate as we
try to rewrite the PTEs inside the already evicted
DPT obj.
TODO: presumably this might also be the reason for the
DPT related display faults under heavy memory pressure,
but I'm still not sure how that would happen as the object
should be pinned by intel_dpt_pin() while in active use by
the display engine...
Cc: stable@vger.kernel.org
Cc: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Fixes: 0dc987b699 ("drm/i915/display: Add smem fallback allocation for dpt")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320090522.9909-2-ville.syrjala@linux.intel.com
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
There are multiple ways in which the GuC load can fail. The driver was
reporting the status register as is, but not everyone can read the
matrix unfiltered. So add decoding of the common error cases.
Also, remove the comment about interrupt based load completion
checking being not recommended. The interrupt was removed from the GuC
firmware some time ago so it is no longer an option anyway. While at
it, also abort the timeout if a known error code is reported. No need
to keep waiting if the GuC has already given up the load.
v2: Fix mis-matched case and confusing 'success' variable (Daniele).
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230316220632.3312218-2-John.C.Harrison@Intel.com
