This patch setup voltage swing before enabling
combo PHY DDI (shared with DSI).
Note that DSI voltage swing programming is for
high speed data buffers. HW automatically handles
the voltage swing for the low power data buffers.
v2: Rebase
v3: Address various review comments related to VSWING
programming (Jani N)
drm/i915/icl: Configure lane sequencing of combo phy transmitter
This patch set the loadgen select and latency optimization for
aux and transmit lanes of combo phy transmitters. It will be
used for MIPI DSI HS operations.
v2: Rebase
v3: Add empty line to make code more legible (Ville).
On skylake we can switch to a high quality scaler mode when only 1 out
of 2 scalers are used, but on GLK and later bit 28 has a different
meaning. Don't set it, and make clear the distinction between
SKL and later PS values.
drm/i915: Replace call to commit_planes_on_crtc with internal update, v2.
drm_atomic_helper_commit_planes_on_crtc calls begin_commit,
then plane_update hooks, then flush_commit. Because we keep our own
visibility tracking through plane_state->visible there's no need to
rely on the atomic hooks for this.
By explicitly writing our own helper, we can update visible planes
as needed, which is useful to make NV12 support work as intended.
Changes since v1:
- Reword commit message. (Matt Roper)
- Rename to intel_update_planes_on_crtc(). (Matt)
drm/i915: Make intel_crtc_disable_planes() use active planes mask.
This will only disable planes we actually had marked as visible in
crtc_state->visible_planes and cleans up intel_crtc_disable_plane()
slightly.
This is also useful for when we start enabling NV12 support for gen11,
in which we will make the separate Y plane visible, but ignore the
Y plane's state.
We need to assume the plane has been visible before, even if no CRTC
is assigned to the plane. This is because when enabling a nv12 plane
on gen11, we will have to enable an extra plane and make it visible
by marking it in crtc_state->active_planes for
intel_update_planes_on_crtc().
Additionally, clear visible flag in intel_plane_atomic_check, in case
we ever hit a bug with visibility. Our code implicitly assumes that
plane_state->visible is only true when crtc and fb are set,
so we will either null deref in intel_fbc_choose_crtc() or
do something bad during the actual commit which cares even more.
Changes since v1:
- Unconditionally clear crtc_state->active_planes as well.
- Reword commit message, since this is now a preparation patch for
NV12 Y / UV plane linking.
While we may not update new_crtc_state, we may clear active_planes
if the new cursor update state will disable the cursor, but we fail
after. If this is immediately followed by a modeset disable, we may
soon not disable the planes correctly when we start depending on
active_planes.
Changes since v1:
- Clarify why we cannot swap crtc_state. (Matt)
Chris Wilson [Thu, 20 Sep 2018 19:59:48 +0000 (20:59 +0100)]
drm/i915/execlists: Onion unwind for logical_ring_init() failure
Fix up the error unwind for logical_ring_init() failing by moving the
cleanup into the callers who own the various bits of state during
initialisation, so we don't forget to free the state allocated by the
caller.
Chris Wilson [Thu, 20 Sep 2018 16:13:43 +0000 (17:13 +0100)]
drm/i915: Park the GPU on module load
Once we have flushed the first request through the system to both load a
context and record the default state; tell the GPU to park and idle
itself, putting itself immediately (hopefully at least) into a
powersaving state, and allowing ourselves to start from known state
after setting up all our bookkeeping.
Chris Wilson [Thu, 20 Sep 2018 14:49:34 +0000 (15:49 +0100)]
drm/i915/selftests: Live tests emit requests and so require rpm
As we emit requests or touch HW directly for some of the live tests, the
requirement is that we hold the rpm wakeref before doing so. We want a
mix of granularity since we will want to test runtime suspend, so try to
mark up only the critical sections where we need rpm for the live test.
Chris Wilson [Wed, 19 Sep 2018 20:54:32 +0000 (21:54 +0100)]
drm/i915/guc: Restore preempt-context across S3/S4
Stolen memory is lost across S4 (hibernate) or S3-RST as it is a portion
of ordinary volatile RAM. As we allocate our rings from stolen, this may
include the rings used for our preempt context and their breadcrumb
instructions. In order to allow preemption following hibernation and
loss of stolen memory, we therefore need to repopulate the instructions
inside the lost ring upon resume. To handle both module load and resume,
we simply defer constructing the ring to first use.
Chris Wilson [Thu, 20 Sep 2018 10:58:09 +0000 (11:58 +0100)]
drm/i915/selftests: Basic stress test for rapid context switching
We need to exercise the HW and submission paths for switching contexts
rapidly to check that features such as execlists' wa_tail are adequate.
Plus it's an interesting baseline latency metric.
v2: Check the initial request for allocation errors
v3: Use finite waits for more robust handling of broken code
drm/i915/psr: Enable AUX-A IO power well on ICL for PSR
PSR requires AUX IO power well to be enabled. This was already in place
for CNL, extend this for ICL too. Not enabling the power well results in
the aux error interrupts when the hardware exits PSR.
Reported-by: Casey G Bowman <casey.g.bowman@intel.com> Reported-by: Jyoti R Yadav <jyoti.r.yadav@intel.com> Cc: Matt Atwood <matthew.s.atwood@intel.com> Cc: Jyoti R Yadav <jyoti.r.yadav@intel.com> Cc: Casey G Bowman <casey.g.bowman@intel.com> Tested-by: Casey G Bowman <casey.g.bowman@intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Reviewed-by: José Roberto de Souza <jose.souza@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180914001822.2503-1-dhinakaran.pandiyan@intel.com
Ville Syrjälä [Mon, 17 Sep 2018 15:15:03 +0000 (18:15 +0300)]
drm/i915/sdvo: Fix multi function encoder stuff
SDVO encoders can have multiple different types of outputs hanging off
them. Currently the code tries to muck around with various is_foo
flags in the encoder to figure out which type its driving. That doesn't
work with atomic and other stuff, so let's nuke those flags and just
look at which type of connector we're actually dealing with.
The is_hdmi we'll need as that's not discoverable via the output flags,
but we'll just move it under the connector.
We'll also move the sdvo fixed mode handling out from the .get_modes()
hook into the sdvo lvds init function so that we can bail out properly
if there is no fixed mode to be found.
Ville Syrjälä [Tue, 18 Sep 2018 13:10:59 +0000 (16:10 +0300)]
drm/i915: Fix logic fumble in rotation vs. ccs check
Smatch reports:
../drivers/gpu/drm/i915/intel_sprite.c:1192 skl_plane_check_fb() warn: was || intended here instead of &&?
Obviously smatch is correct here since we're trying to check if we're
using either of the ccs modifiers. Since we now have is_ccs_modifier()
let's use it to fix this.
Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: José Roberto de Souza <jose.souza@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Fixes: e21c2d331018 ("drm/i915: Move skl plane fb related checks into a better place") Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180918131059.793-1-ville.syrjala@linux.intel.com Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Ville Syrjälä [Mon, 17 Sep 2018 17:14:14 +0000 (20:14 +0300)]
drm/i915: Replace some PAGE_SHIFTs with I915_GTT_PAGE_SIZE
Clean up some cases where we're dealing with GTT pages instead of
system pages to use I915_GTT_PAGE_SIZE instead of PAGE_SHIT. So
just replace the the shifts with mul/div as appropriate. These
are the easy ones, the rest probably need some actual thought.
No real changes in the generated asm. Only gen8_ppgtt_insert_4lvl()
was affected as gcc decided to do the following change:
- be9: 89 d9 mov %ebx,%ecx
- beb: c1 e1 0c shl $0xc,%ecx
- bee: 48 63 c9 movslq %ecx,%rcx
+ be9: 48 63 cb movslq %ebx,%rcx
+ bec: 48 c1 e1 0c shl $0xc,%rcx
and that then shifted a bunch of the offset by one byte. I presume
the sign extensions in the asm are due to integer promotions from
u16 etc. Hopefully someone has confirmed that those don't end up
doing the wrong thing for us.
Chris Wilson [Tue, 11 Sep 2018 11:57:47 +0000 (12:57 +0100)]
drm/i915: Limit number of capture objects
If we fail to allocate an array for a large number of user requested
capture objects, reduce the array size and try to grab at least some of
the objects!
Chris Wilson [Fri, 14 Sep 2018 08:00:17 +0000 (09:00 +0100)]
drm/i915/execlists: Reset CSB pointers on canceling requests (wedging)
The prior assumption was that we did not need to reset the CSB on
wedging when cancelling the outstanding requests as it would be cleaned
up in the subsequent reset prior to restarting the GPU. However, what
was not accounted for was that in preparing for the reset, we would try
to process the outstanding CSB entries. If the GPU happened to complete
a CS event just as we were performing the cancellation of requests, that
event would be kept in the CSB until the reset -- but our bookkeeping
was cleared, causing confusion when trying to complete the CS event.
v2: Use a sanitize on unwedge to avoid interfering with eio suspend
(where we intentionally disable GPU reset).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107925 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180914080017.30308-3-chris@chris-wilson.co.uk
Chris Wilson [Fri, 14 Sep 2018 12:40:07 +0000 (13:40 +0100)]
drm/i915: Include fence-hint for timeout warning
If an asynchronous wait on a foriegn fence, we print a warning
indicating which fence was not signaled. As i915_sw_fences become more
common, include the debug hint (the symbol-name of the target) to help
identify the waiter. E.g.
[ 31.968144] Asynchronous wait on fence sw_sync:gem_eio:1 timed out (hint:submit_notify [i915])
We also want to downgrade from a warning to a notice (normal but
significant condition) as the timeout is imposed and controlled by the
caller (i.e. it is deliberate) and can be provoked by userspace.
Chris Wilson [Fri, 14 Sep 2018 12:35:04 +0000 (13:35 +0100)]
drm/i915/execlists: Use coherent writes into the context image
That we use a WB mapping for updating the RING_TAIL register inside the
context image even on !llc machines has been a source of consternation
for every reader. It appears to work on bsw+, but it may just have been
that we have been incredibly bad at detecting the errors.
v2: With extra enthusiasm.
v3: Drop force of map type for pinned default_state as by the time we
pin it, the map type is always WB and doesn't conflict with the earlier
use by ce->state.
v4: Transfer engine->default_state from MAP_WC to MAP_WB on creation so
we do not need the MAP_FORCE littered around the backends
Chris Wilson [Fri, 14 Sep 2018 12:35:03 +0000 (13:35 +0100)]
drm/i915: Check engine->default_state mapping on module load
Check we can indeed acquire a WB mapping of the context image on module
load. Later this will give us the opportunity to validate that we can
switch from WC to WB as required.
Chris Wilson [Fri, 14 Sep 2018 12:35:02 +0000 (13:35 +0100)]
drm/i915/execlists: Delay updating ring register state after resume
Now that we reload both RING_HEAD and RING_TAIL when rebinding the
context, we do not need to scrub those registers immediately on resume.
v2: Handle the perma-pinned contexts.
v3: Set RING_TAIL on context-pin so that we always have known state in
the context image for the ring registers and all parties have similar
code (ripe for refactoring).
Chris Wilson [Fri, 14 Sep 2018 08:00:16 +0000 (09:00 +0100)]
drm/i915: Flush the tasklet when checking for idle
In order to reduce latency when checking for idle we kick the tasklet
directly. Sometimes this is not enough as it is queued on another cpu
and so to improve the accuracy of this idle-check (and so to reduce
latency overall by avoiding another pass, or worse declaring a timeout!)
wait for the tasklet to complete.
References: https://bugs.freedesktop.org/show_bug.cgi?id=107916 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Cc: Michel Thierry <michel.thierry@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180914080017.30308-2-chris@chris-wilson.co.uk
Chris Wilson [Fri, 14 Sep 2018 08:00:15 +0000 (09:00 +0100)]
drm/i915: Limit the backpressure for i915_request allocation
If we try and fail to allocate a i915_request, we apply some
backpressure on the clients to throttle the memory allocations coming
from i915.ko. Currently, we wait until completely idle, but this is far
too heavy and leads to some situations where the only escape is to
declare a client hung and reset the GPU. The intent is to only ratelimit
the allocation requests and to allow ourselves to recycle requests and
memory from any long queues built up by a client hog.
Although the system memory is inherently a global resources, we don't
want to overly penalize an unlucky client to pay the price of reaping a
hog. To reduce the influence of one client on another, we can instead of
waiting for the entire GPU to idle, impose a barrier on the local client.
(One end goal for request allocation is for scalability to many
concurrent allocators; simultaneous execbufs.)
To prevent ourselves from getting caught out by long running requests
(requests that may never finish without userspace intervention, whom we
are blocking) we need to impose a finite timeout, ideally shorter than
hangcheck. A long time ago Paul McKenney suggested that RCU users should
ratelimit themselves using judicious use of cond_synchronize_rcu(). This
gives us the opportunity to reduce our indefinite wait for the GPU to
idle to a wait for the RCU grace period of the previous allocation along
this timeline to expire, satisfying both the local and finite properties
we desire for our ratelimiting.
There are still a few global steps (reclaim not least amongst those!)
when we exhaust the immediate slab pool, at least now the wait is itself
decoupled from struct_mutex for our glorious highly parallel future!
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106680 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180914080017.30308-1-chris@chris-wilson.co.uk
Mahesh Kumar [Fri, 24 Aug 2018 09:32:25 +0000 (15:02 +0530)]
drm/i915/kbl+: Enable IPC only for symmetric memory configurations
IPC may cause underflows if not used with dual channel symmetric
memory configuration. Disable IPC for non symmetric configurations in
affected platforms.
Display WA #1141
Changes Since V1:
- Re-arrange the code.
- update wrapper to return if memory is symmetric (Rodrigo)
Mahesh Kumar [Fri, 31 Aug 2018 11:09:42 +0000 (16:39 +0530)]
drm/i915: Implement 16GB dimm wa for latency level-0
Memory with 16GB dimms require an increase of 1us in level-0 latency.
This patch implements the same.
Bspec: 4381
changes since V1:
- s/memdev_info/dram_info
- make skl_is_16gb_dimm pure function
Changes since V2:
- make is_16gb_dimm more generic
- rebase
Changes since V3:
- Simplify condition (Maarten)
Mahesh Kumar [Fri, 24 Aug 2018 09:32:22 +0000 (15:02 +0530)]
drm/i915/skl+: Decode memory bandwidth and parameters
This patch adds support to decode system memory bandwidth and other
parameters for skylake and Gen9+ platforms, which will be used for
arbitrated display memory bandwidth calculation in GEN9 based
platforms and WM latency level-0 Work-around calculation on GEN9+.
Changes Since V1:
- s/memdev_info/dram_info
- create a struct to hold channel info
Changes Since V2:
- rewrite code to adhere i915 coding style
- not valid for GLK
Mahesh Kumar [Fri, 24 Aug 2018 09:32:21 +0000 (15:02 +0530)]
drm/i915/bxt: Decode memory bandwidth and parameters
This patch adds support to decode system memory bandwidth and other
parameters for broxton platform, which will be used for arbitrated
display memory bandwidth calculation in GEN9 based platforms and
WM latency level-0 Work-around calculation on GEN9+ platforms.
Changes since V1:
- s/memdev_info/dram_info
Changes since V2:
- Adhere to i915 coding style (Rodrigo)
Ville Syrjälä [Wed, 12 Sep 2018 18:04:43 +0000 (21:04 +0300)]
drm/i915: Fix a potential integer overflow with framebuffers extending past 4 GiB
If we have framebuffers that are >= 4GiB in size we will overflow
the fb size check in intel_fill_fb_info().
Currently that is only possible with NV12 and CCS as offsets[1]
may be anything between 0 and 0xffffffff. offsets[0] is currently
required to be 0 so we can't hit the overflow with any single
plane format (thanks to max fb size of 8kx8k and max stride of
32 KiB).
In the future we may allow almost any framebuffer to exceed 4GiB
in size so we really should fix the overflow. Not that the overflow
is particularly dangerous. It's mostly just a sanity check against
insane userspace. The display engine can't write to memory anyway
so I suppose in the worst case we might anger the hw by attempting
scanout past the end of the ggtt, or we might scan out some data
that we're not supposed to see from other parts of the ggtt.
Note that triggering this overflow depends on the driver
aligning the fb height to the next tile boundary to push the
calculated size above 4GiB. With linear buffers the effective
tile height is one so that never happens, and the core already
has a check for 32bit overflow of offsets[]+pitches[]*height.
We can remove the update-via-batch-buffer code path, which is basically an
effective duplicate of update-via-context-image path, if we notice that
after we have idled the GPU, we can update the context image even of the
kernel context directly. (Update-via-batch-buffer path existed only to
solve the problem of how to update the kernel context image.)
Only additional thing needed is to activate the edited configuration by
sending one empty request down the pipe. This accomplishes context restore
of the updated kernel context and so the OA configuration gets written out
to it's control registers.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Cc: Matthew Auld <matthew.auld@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20180912152930.28237-1-tvrtko.ursulin@linux.intel.com
Ville Syrjälä [Fri, 7 Sep 2018 15:24:11 +0000 (18:24 +0300)]
drm/i915: Move display w/a #1175
Move the display w/a #1175 to a better place. That place
being the new skl+ specific plane->check() hook. This leaves
the skl_check_plane_surface() stuff to deal with the gtt offset
and src coordinate stuff as originally envisioned.
Ville Syrjälä [Fri, 7 Sep 2018 15:24:10 +0000 (18:24 +0300)]
drm/i915: Move skl plane fb related checks into a better place
Move the skl+ specific framebuffer related checks from
intel_plane_atomic_check_with_state() into a new function
(skl_plane_check_fb()) which we'll simply call from the skl
plane->check() hook.
v2: Split out the Y/Yf+CCS vs. interlaced change (José)
Split up intel_check_primary_plane() and intel_check_sprite_plane()
into per-platform variants. This way we can get a unified behaviour
between the SKL universal planes, and we stop checking for non-SKL
specific scaling limits for the "sprite" planes. And we now get
a natural place where to add more plarform specific checks.
v2: Split the .check_plane() calling convention change out (José)
Ville Syrjälä [Fri, 7 Sep 2018 15:24:08 +0000 (18:24 +0300)]
drm/i915: Nuke plane->can_scale/min_downscale
We can easily calculate the plane can_scale/min_downscale on demand.
And later on we'll probably want to start calculating these dynamically
based on the cdclk just as skl already does.
Ville Syrjälä [Fri, 7 Sep 2018 15:24:06 +0000 (18:24 +0300)]
drm/i915: Store ggtt_view in plane_state
Stash the gtt_view structure into the plane state. This will become
useful when we do GTT remapping as the gtt_view will not come directly
from the fb anymore.
Ville Syrjälä [Tue, 11 Sep 2018 15:01:39 +0000 (18:01 +0300)]
drm/i915: Store the final plane stride in plane_state
Let's store the final plane stride in the plane state. This avoids
having to pick between the normal vs. rotated stride during hardware
programming. And once we get GTT remapping the plane stride will
no longer match the fb stride so we'll need a place to store it
anyway.
v2: Keep checking fb->pitches[0] for cursor as later on we won't
populate plane_state->color_plane[0].stride for invisible planes
and we have been checking the cursor fb stride even for invisible
planes
v3: s/betwen/between in commit msg (José)
v4: Check color_plane[0].stride instead of fb->pitches[0] in
the skl_check_main_surface() X-tiling kludge
Ville Syrjälä [Fri, 7 Sep 2018 15:24:04 +0000 (18:24 +0300)]
drm/i915: Rename the plane_state->main/aux to plane_state->color_plane[]
Make the main/aux surface stuff a bit more generic by using an array
of structures. This will allow us to deal with both the main and aux
surfaces with common code.
Ville Syrjälä [Fri, 7 Sep 2018 15:24:03 +0000 (18:24 +0300)]
drm/i915: Use pipe A primary plane .max_stride() as the global stride limit
Let's assume that the primary plane for pipe A has the highest max
stride of all planes, and we'll use that as the global limit when
creating a new framebuffer.
Ville Syrjälä [Fri, 7 Sep 2018 15:24:02 +0000 (18:24 +0300)]
drm/i915: Add .max_stride() plane hook
Each plane may have different stride limitations. Let's add a new
plane function to retutn the maximum stride for each plane. There's
going to be some use for this outside the .atomic_check() stuff hence
the separate hook.
Ville Syrjälä [Fri, 7 Sep 2018 15:24:01 +0000 (18:24 +0300)]
drm/i915: s/tile_offset/aligned_offset/ etc.
Rename some of the tile_offset() functions to aligned_offset() since
they operate on both linear and tiled functions. And we'll include
_plane_ in the name of all the variants that take a plane state.
Should make it more clear which function to use where.
Chris Wilson [Wed, 12 Sep 2018 10:11:33 +0000 (11:11 +0100)]
drm/i915: Reorder execobject[] to insert non-48b objects into the low 4G
If the caller supplies more than 4G of objects and than one that has to
be in the low 4G, it is possible for the low 4G to be full before we
attempt to find room for the last object that must be there. As we don't
reorder the two types, every pass hits the same problem and we fail with
ENOSPC. However, if we impose a little bit of ordering between the two
classes of objects, on the second pass we will be able to fit the
special object as we do it first. For setups that only use !48b objects,
we now reverse the order between passes, hopefully making the subsequent
passes more likely to succeed given that we are trying a different
order (rather than repeating the previous pass!)
v2: Quick one line explanation for the relative priorities given to
reservations.
Chris Wilson [Mon, 10 Sep 2018 13:08:08 +0000 (14:08 +0100)]
drm/i915/ringbuffer: Reload PDs harder on byt/bcs
Baytrail takes a little more convincing that it needs to actually reload
its Page Directoy (ppGTT) before the context switch, so repeat it until
it gets the message. Once again the arbitrary values here are
empirically derived.
Chris Wilson [Tue, 11 Sep 2018 13:22:06 +0000 (14:22 +0100)]
drm/i915: Nuke struct_mutex from context_setparam
Userspace should be free to race against itself and shoot itself in
the foot if it so desires to adjust a parameter at the same time as
submitting a batch to that context. As such, the struct_mutex in context
setparam is only being used to serialise userspace against itself and
not for any protection of internal structs and so is superfluous.
v2: Separate user_flags from internal flags to reduce chance of
interference; and use locked bit ops for user updates.
Chris Wilson [Thu, 6 Sep 2018 19:01:44 +0000 (20:01 +0100)]
drm/i915/overlay: Use the ioctl parameters directly
The user parameters to put_image are not copied back to userspace
(DRM_IOW), and so we can modify the ioctl parameters (having already been
copied to a temporary kernel struct) directly and use those in place,
avoiding another temporary malloc and lots of manual copying.
Chris Wilson [Thu, 6 Sep 2018 19:01:43 +0000 (20:01 +0100)]
drm/i915/overlay: Allocate physical registers from stolen
Given that we are now reasonably confident in our ability to detect and
reserve the stolen memory (physical memory reserved for graphics by the
BIOS) for ourselves on most machines, we can put it to use. In this
case, we need a page to hold the overlay registers.
On an i915g running MythTv, H Buus noticed that
commit 6a2c4232ece145d8b5a8f95f767bd6d0d2d2f2bb
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Tue Nov 4 04:51:40 2014 -0800
drm/i915: Make the physical object coherent with GTT
introduced stuttering into his video playback. After discarding the
likely suspect of it being the physical cursor updates, we were left
with the use of the phys object for the overlay. And lo, if we
completely avoid using the phys object (allocated just once on module
load!) by switching to stolen memory, the stuttering goes away.
For lack of a better explanation, claim victory and kill two birds with
one stone.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107600 Fixes: 6a2c4232ece1 ("drm/i915: Make the physical object coherent with GTT") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180906190144.1272-1-chris@chris-wilson.co.uk
drm/i915/chv: Update csc coefficient matrix during modeset
During modeset, previously configured csc coefficient matrix,if any, will
not persist. This can result in blank screen as csc mode will be programmed
while loading LUT but csc coefficient matrix remains unprogrammed.
Michal Wajdeczko [Mon, 10 Sep 2018 10:41:49 +0000 (10:41 +0000)]
drm/i915/guc: Update GuC power domain states
We should update GuC power domain states also when GuC submission
is disabled, otherwise GuC might complain or ignore our requests.
This seems to be required for all currently released GuC firmwares.
v2: it is only needed by pre-Gen11 firmwares
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: John Spotswood <john.a.spotswood@intel.com> Cc: Anusha Srivatsa <anusha.srivatsa@intel.com> Cc: Tomasz Lis <tomasz.lis@intel.com> Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Reviewed-by: Anusha Srivatsa <anusha.srivatsa@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20180910104150.101752-1-michal.wajdeczko@intel.com
Chris Wilson [Fri, 7 Sep 2018 11:28:51 +0000 (12:28 +0100)]
drm/i915: Missed interrupt simulation is no more, tell the world
Using the guc, we cannot disable the user interrupt generation as we use
it for driving submission. And from Icelake, we no longer have the
ability to individually mask interrupt generation from each engine,
disabling our ability to fake missed interrupts.
In both cases, report back to userspace that the missed interrupt
generator is no longer available.
Chris Wilson [Tue, 4 Sep 2018 15:31:17 +0000 (16:31 +0100)]
drm/i915: Reduce context HW ID lifetime
Future gen reduce the number of bits we will have available to
differentiate between contexts, so reduce the lifetime of the ID
assignment from that of the context to its current active cycle (i.e.
only while it is pinned for use by the HW, will it have a constant ID).
This means that instead of a max of 2k allocated contexts (worst case
before fun with bit twiddling), we instead have a limit of 2k in flight
contexts (minus a few that have been pinned by the kernel or by perf).
To reduce the number of contexts id we require, we allocate a context id
on first and mark it as pinned for as long as the GEM context itself is,
that is we keep it pinned it while active on each engine. If we exhaust
our context id space, then we try to reclaim an id from an idle context.
In the extreme case where all context ids are pinned by active contexts,
we force the system to idle in order to recover ids.
We cannot reduce the scope of an HW-ID to an engine (allowing the same
gem_context to have different ids on each engine) as in the future we
will need to preassign an id before we know which engine the
context is being executed on.
v2: Improved commentary (Tvrtko) [I tried at least]
References: https://bugs.freedesktop.org/show_bug.cgi?id=107788 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Michel Thierry <michel.thierry@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180904153117.3907-1-chris@chris-wilson.co.uk
Chris Wilson [Tue, 4 Sep 2018 16:29:02 +0000 (17:29 +0100)]
drm/i915: Be defensive and don't assume PSR has any commit to sync against
If the previous modeset commit has completed and is no longer part of
the crtc state, skip waiting for it.
Ville pointed out that, in fact, the commit is never removed after a
modeset so the only way we could see a NULL here should be if there was
never a commit attached. Nevertheless, we have the evidence it can be
NULL and it has been defended against elsewhere, for example commit 93313538c153 ("drm/i915: Pass idle crtc_state to intel_dp_sink_crc").
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107792 Fixes: c44301fce614 ("drm/i915: Allow control of PSR at runtime through debugfs, v6") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180904162902.2578-1-chris@chris-wilson.co.uk
Chris Wilson [Tue, 4 Sep 2018 13:12:07 +0000 (14:12 +0100)]
drm/i915: Pull intel_uncore_arm_unclaimed_mmio_detection() under the spinlock
Elsewhere we manipulate uncore.unclaimed_mmio_check and
i915_param.mmio_debug under the irq lock (e.g. preserving the current
value across a user forcewake grab), but do not protect the manipulation
inside intel_uncore_arm_unclaimed_mmio_detection() from concurrent
access, even from itself. This is an issue as we do call
arm_unclaimed_mmio_detection from multiple threads without coordination.
There are two issues with the current RPCS programming for Icelake:
Expansion of the slice count bitfield has been missed, as well as the
required programming workaround for the subslice count bitfield size
limitation.
1)
Bitfield width for configuring the active slice count has grown so we need
to program the GEN8_R_PWR_CLK_STATE accordingly.
Current code was always requesting eight times the number of slices (due
writing to a bitfield starting three bits higher than it should). These
requests were luckily a) capped by the hardware to the available number of
slices, and b) we haven't yet exported the code to ask for reduced slice
configurations.
Due both of the above there was no impact from this incorrect programming
but we should still fix it.
2)
Due subslice count bitfield being only three bits wide and furthermore
capped to a maximum documented value of four, special programming
workaround is needed to enable more than four subslices.
With this programming driver has to consider the GT configuration as
2x4x8, while the hardware internally translates this to 1x8x8.
A limitation stemming from this is that either a subslice count between
one and four can be selected, or a subslice count equaling the total
number of subslices in all selected slices. In other words, odd subslice
counts greater than four are impossible, as are odd subslice counts
greater than a single slice subslice count.
This also had no impact in the current code base due breakage from 1)
always reqesting more than one slice.
While fixing this we also add some asserts to flag up any future bitfield
overflows.
v2:
* Use a local in all branches for clarity. (Lionel)
Chris Wilson [Tue, 4 Sep 2018 06:38:02 +0000 (07:38 +0100)]
drm/i915/ringbuffer: Move double invalidate to after pd flush
Continuing the fun of trying to find exactly the delay that is
sufficient to ensure that the page directory is fully loaded between
context switches, move the extra flush added in commit 70b73f9ac113
("drm/i915/ringbuffer: Delay after invalidating gen6+ xcs") to just
after we flush the pd. Entirely based on the empirical data of running
failing tests in a loop until we survive a day (before the mtbf is 10-30
minutes).
Chris Wilson [Tue, 4 Sep 2018 11:17:32 +0000 (12:17 +0100)]
drm/i915: Double check we didn't miss an unclaimed register access
Currently, if the user has enabled mmio-debug around each register
access, we presume that we have then checked them all. However, it is
still possible through omission (raw register access) or external
interaction that the unclaimed access was not highlighted.
Chris Wilson [Mon, 3 Sep 2018 15:23:04 +0000 (16:23 +0100)]
drm/i915: Use a cached mapping for the physical HWS
Older gen use a physical address for the hardware status page, for which
we use cache-coherent writes. As the writes are into the cpu cache, we use
a normal WB mapped page to read the HWS, used for our seqno tracking.
Anecdotally, I observed lost breadcrumbs writes into the HWS on i965gm,
which so far have not reoccurred with this patch. How reliable that
evidence is remains to be seen.
v2: Explicitly pass the expected physical address to the hw
v3: Also remember the wild writes we once had for HWS above 4G.
Chris Wilson [Mon, 3 Sep 2018 15:02:16 +0000 (16:02 +0100)]
drm/i915: Fix up FORCE_GPU_RELOC (debug) to flush CPU write domains
We currently assert that if the target is in a CPU write domain, we use
a CPU reloc path rather than the GPU reloc path. However, we have a debug
override to force the GPU path and that unfortunately hits the assert.
Include the async clflush under the debug option to ensure correct
behaviour even when debugging, and strict when not.
Chris Wilson [Mon, 3 Sep 2018 08:33:37 +0000 (09:33 +0100)]
drm/i915: Forcibly flush unwanted requests in drop-caches
Add a mode to debugfs/drop-caches to flush unwanted requests off the GPU
(by wedging the device and resetting). This is very useful if a test
terminated leaving a long queue of hanging batches that would ordinarily
require a round trip through hangcheck for each.
It reduces the inter-test operation to just a write into drop-caches to
reset driver/GPU state between tests.
Chris Wilson [Mon, 3 Sep 2018 08:33:36 +0000 (09:33 +0100)]
drm/i915: Early rejection of buffer allocations larger than RAM
We currently try to pin and allocate the whole buffer at a time. If that
object is larger than RAM, we will try to pin the whole of physical
memory, force the machine into oom, and then still fail the allocation.
If the request is obviously too large, error out early. We opt to do
this in the backend to make it easy to use alternate paths that do not
require the entire object pinned, or may easily handle proxy objects
that are larger than physical memory.
Chris Wilson [Mon, 3 Sep 2018 08:33:35 +0000 (09:33 +0100)]
drm/i915: Force the slow path after a user-write error
If we fail to write the user relocation back when it is changed, force
ourselves to take the slow relocation path where we can handle faults in
the write path. There is still an element of dubiousness as having
patched up the batch to use the correct offset, it no longer matches the
presumed_offset in the relocation, so a second pass may miss any changes
in layout.
Chris Wilson [Mon, 3 Sep 2018 08:33:34 +0000 (09:33 +0100)]
drm/i915: Flag any possible writes for a GTT fault
We do not explicitly mark the PTE for the user's GTT mmap as being
wrprotect, so we don't get a refault when we would need to change a
read-only mmapping into read-write. As such, we must presume that if the
vma has PROT_WRITE it may be written to, although this is supposed to be
indicated by set-domain there are cases (e.g. after swap) where
userspace may not be aware of the implicit domain change.
Chris Wilson [Mon, 3 Sep 2018 08:33:33 +0000 (09:33 +0100)]
drm/i915: Do a full device reset after being wedged
We only call unset_wedged on the global reset path (since it's a global
operation), so if we are terminally wedged and wish to reset, take the
full device reset path rather than the quicker individual engine resets.
Chris Wilson [Sat, 1 Sep 2018 09:24:51 +0000 (10:24 +0100)]
drm/i915: Determine uses-full-ppgtt from context for execbuf
Rather than inspect the global module parameter for whether full-ppgtt
maybe enabled, we can inspect the context directly as to whether it has
its own vm.
Tvrtko Ursulin [Fri, 31 Aug 2018 14:36:43 +0000 (15:36 +0100)]
drm/i915: Explicitly mark Global GTT address spaces
So far we have been relying on vm->file pointer being NULL to declare
something GGTT.
This has the unfortunate consequence that the default kernel context is
also declared GGTT and interferes with the following patch which wants to
instantiate VMA's and execute requests against the kernel context.
Change the is_ggtt test to use an explicit flag in struct address_space to
solve this issue.
Note that the bit used is free since there is an alignment hole in the
struct.
Imre Deak [Fri, 31 Aug 2018 17:47:39 +0000 (20:47 +0300)]
drm/i915/dp_mst: Fix enabling pipe clock for all streams
commit afb2c4437dae ("drm/i915/ddi: Push pipe clock enabling to encoders")
inadvertently stopped enabling the pipe clock for any DP-MST stream
after the first one. It also rearranged the pipe clock enabling wrt.
initial MST payload allocation step (which may or may not be a
problem, but it's contrary to the spec.).
Fix things by making the above commit truly a non-functional change.
Fixes: afb2c4437dae ("drm/i915/ddi: Push pipe clock enabling to encoders")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107365 Reported-by: Lyude Paul <lyude@redhat.com> Reported-by: dmummenschanz@web.de Tested-by: dmummenschanz@web.de Tested-by: Lyude Paul <lyude@redhat.com> Cc: Lyude Paul <lyude@redhat.com> Cc: dmummenschanz@web.de Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180831174739.30387-1-imre.deak@intel.com
Jyoti Yadav [Fri, 31 Aug 2018 06:00:23 +0000 (02:00 -0400)]
drm/i915/intel_csr.c Fix DMC FW Loading issue on ICL.
This patch resolves the DMC FW loading issue.
Earlier DMC FW package have only one DMC FW for one stepping. But as such
there is no such restriction from Package side.
For ICL icl_dmc_ver1_07.bin binary package has DMC FW for 2 steppings.
So while reading the dmc_offset from package header, for 1st stepping
offset used to come 0x0 and was working fine till now.
But for second stepping and other steppings, offset is non zero number
and is in dwords. So we need to convert into bytes to fetch correct DMC
FW from correct place.
v2 : Added check for DMC FW max size for various gen. (Imre Deak)
v3 : Corrected naming convention for various gen. (Imre Deak)
v4 : Initialized max_fw_size to 0
v5 : Corrected DMC FW MAX_SIZE for various gen. (Imre Deak)
v6 : Fixed the typo issues.
Chris Wilson [Thu, 30 Aug 2018 13:48:06 +0000 (14:48 +0100)]
drm/i915/selftests: Add a simple exerciser for suspend/hibernate
Although we cannot do a full system-level test of suspend/hibernate from
deep with the kernel selftests, we can exercise the GEM subsystem in
isolation and simulate the external effects (such as losing stolen
contents and trashing the register state).
v2: Don't forget to hold rpm
v3: Suspend the GTT mappings, and more rpm!
References: https://bugs.freedesktop.org/show_bug.cgi?id=96526
References: 5ab57c702069 ("drm/i915: Flush logical context image out to memory upon suspend") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Jakub Bartmiński <jakub.bartminski@intel.com> Cc: Matthew Auld <matthew.william.auld@gmail.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by: Jakub Bartmiński <jakub.bartminski@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180830134806.21939-1-chris@chris-wilson.co.uk
Chris Wilson [Fri, 17 Aug 2018 08:24:05 +0000 (09:24 +0100)]
drm/i915: Keep physical cursors pinned while in use
The optimisation inherent in commit 6a2c4232ece1 ("drm/i915: Make the
physical object coherent with GTT") relies on that once we allocated a
cursor we would have coherent, zero overhead access to the scanout plane
holding the cursor. That is we could then do the very frequent cursor
updates X enjoys with no indirection or kernel involvement. However,
that all hinges on the GGTT mmap of the cursor being pinned and not
require refaulting on each access -- handling such a page fault likely
requires the busy GGTT to be rearranged causing a stall. A very simple
fix is then to handle the physical cursor exactly like other cursors and
keep its vma pinned while active.
Chris Wilson [Thu, 30 Aug 2018 16:10:42 +0000 (17:10 +0100)]
drm/i915/ringbuffer: Delay after invalidating gen6+ xcs
During stress testing of full-ppgtt (on Baytrail at least), we found
that the invalidation around a context/mm switch was insufficient (writes
would go astray). Adding a second MI_FLUSH_DW barrier prevents this, but
it is unclear as to whether this is merely a delaying tactic or if it is
truly serialising with the TLB invalidation. Either way, it is
empirically required.
v2: Avoid the loop for readability;
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107715
References: https://bugs.freedesktop.org/show_bug.cgi?id=107759 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Cc: Matthew Auld <matthew.william.auld@gmail.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180830161042.29193-1-chris@chris-wilson.co.uk
Chris Wilson [Tue, 28 Aug 2018 15:27:02 +0000 (16:27 +0100)]
drm/i915/execlists: Flush tasklet directly from reset-finish
On finishing the reset, the intention is to restart the GPU before we
relinquish the forcewake taken to handle the reset - the goal being the
GPU reloads a context before it is allowed to sleep. For this purpose,
we used tasklet_flush() which although it accomplished the goal of
restarting the GPU, carried with it a sting in its tail: it cleared the
TASKLET_STATE_SCHED bit. This meant that if another CPU queued a new
request to this engine, we would clear the flag and later attempt to
requeue the tasklet on the local CPU, breaking the per-cpu softirq
lists.
Remove the dangerous tasklet_kill() and just run the tasklet func
directly as we know it is safe to do so (the tasklets are internally
locked to allow mixed usage from direct submission).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Michel Thierry <michel.thierry@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180828152702.27536-1-chris@chris-wilson.co.uk
projects / mirror_ubuntu-hirsute-kernel.git / log