A new Kconfig option CONFIG_DRM_I915_REQUEST_TIMEOUT is added, defaulting
to 20s, and this timeout is applied to all users contexts using the
previously added watchdog facility.
Result of this is that any user submission will simply fail after this
timeout, either causing a reset (for non-preemptable), or incomplete
results.
This can have an effect that workloads which used to work fine will
suddenly start failing. Even workloads comprised of short batches but in
long dependency chains can be terminated.
And because of lack of agreement on usefulness and safety of fence error
propagation this partial execution can be invisible to userspace even if
it is "listening" to returned fence status.
Another interaction is with hangcheck where care needs to be taken timeout
is not set lower or close to three times the heartbeat interval. Otherwise
a hang in any application can cause complete termination of all
submissions from unrelated clients. Any users modifying the per engine
heartbeat intervals therefore need to be aware of this potential denial of
service to avoid inadvertently enabling it.
Given all this I am personally not convinced the scheme is a good idea.
Intuitively it feels object importers would be better positioned to
enforce the time they are willing to wait for something to complete.
v2:
* Improved commit message and Kconfig text.
* Pull in some helper code from patch which got dropped.
v3:
* Bump timeout to 20s to see if it helps Tigerlake.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324121335.2307063-7-tvrtko.ursulin@linux.intel.com
Prepares the plumbing for setting request/fence expiration time. All code
is put in place but is never activated due yet missing ability to actually
configure the timer.
Outline of the basic operation:
A timer is started when request is ready for execution. If the request
completes (retires) before the timer fires, timer is cancelled and nothing
further happens.
If the timer fires request is added to a lockless list and worker queued.
Purpose of this is twofold: a) It allows request cancellation from a more
friendly context and b) coalesces multiple expirations into a single event
of consuming the list.
Worker locklessly consumes the list of expired requests and cancels them
all using previous added i915_request_cancel().
Associated timeout value is stored in rq->context.watchdog.timeout_us.
v2:
* Log expiration.
v3:
* Include more information about user timeline in the log message.
v4:
* Remove obsolete comment and fix formatting. (Matt)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324121335.2307063-6-tvrtko.ursulin@linux.intel.com
Disallow sentinel requests follow previous sentinels to make request
cancellation work better when faced with a chain of requests which have
all been marked as in error.
Because in cases where we end up with a stream of cancelled requests we
want to turn off request coalescing so they each will get individually
skipped by the execlists_schedule_in (which is called per ELSP port, not
per request).
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
[danvet: Fix typo in the commit message that Matthew spotted.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324121335.2307063-4-tvrtko.ursulin@linux.intel.com
Currently, we cancel outstanding requests within a context when the
context is closed. We may also want to cancel individual requests using
the same graceful preemption mechanism.
v2 (Tvrtko):
* Cancel waiters carefully considering no timeline lock and RCU.
* Fixed selftests.
v3 (Tvrtko):
* Remove error propagation to waiters for now.
v4 (Tvrtko):
* Rebase for extracted i915_request_active_engine. (Matt)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
[danvet: Resolve conflict because intel_engine_flush_scheduler is
still called intel_engine_flush_submission]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324121335.2307063-3-tvrtko.ursulin@linux.intel.com
- DisplayPort LTTPR fixes around link training and limiting it
according to supported spec version. (Imre)
- Fix enabled_planes bitmask to really represent only logically
enabled planes (Ville).
- Fix DSS CTL registers for ICL DSI transcoders (Jani)
- Fix the GT fence revocation runtime PM logic. (Imre)
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YFxYdrjqeUtSu+3p@intel.com
The i915 driver has its own tasklet interface which was overseen in the
tasklet rework. __tasklet_disable_sync_once() is a wrapper around
tasklet_unlock_wait(). tasklet_unlock_wait() might sleep, but the i915
wrappers invokes it from non-preemtible contexts with bottom halves disabled.
Use tasklet_unlock_spin_wait() instead which can be invoked from
non-preemptible contexts.
Fixes: da04474740 ("tasklets: Replace spin wait in tasklet_unlock_wait()")
Reported-by: kernel test robot <oliver.sang@intel.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20210323092221.awq7g5b2muzypjw3@flow
When CONFIG_OF is disabled, building with 'make W=1' produces warnings
about out of bounds array access:
drivers/gpu/drm/imx/imx-ldb.c: In function 'imx_ldb_set_clock.constprop':
drivers/gpu/drm/imx/imx-ldb.c:186:8: error: array subscript -22 is below array bounds of 'struct clk *[4]' [-Werror=array-bounds]
Add an error check before the index is used, which helps with the
warning, as well as any possible other error condition that may be
triggered at runtime.
The warning could be fixed by adding a Kconfig depedency on CONFIG_OF,
but Liu Ying points out that the driver may hit the out-of-bounds
problem at runtime anyway.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Liu Ying <victor.liu@nxp.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
LDB channel1 should be registered if it is the only channel to be used.
Without this patch, imx_ldb_bind() would skip registering LDB channel1
if LDB channel0 is not used, no matter LDB channel1 needs to be used or
not.
Fixes: 8767f4711b (drm/imx: imx-ldb: move initialization into probe)
Signed-off-by: Liu Ying <victor.liu@nxp.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Put DRM device on initialization failure path rather than directly
return error code.
Fixes: a67d5088ce ("drm/imx: drop explicit drm_mode_config_cleanup")
Signed-off-by: Pan Bian <bianpan2016@163.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
While Kepler does technically support 256x256 cursors, it turns out that
Kepler actually has some additional requirements for scanout surfaces that
we're not enforcing correctly, which aren't present on Maxwell and later.
Cursor surfaces must always use small pages (4K), and overlay surfaces must
always use large pages (128K).
Fixing this correctly though will take a bit more work: as we'll need to
add some code in prepare_fb() to move cursor FBs in large pages to small
pages, and vice-versa for overlay FBs. So until we have the time to do
that, just limit cursor surfaces to 128x128 - a size small enough to always
default to small pages.
This means small ovlys are still broken on Kepler, but it is extremely
unlikely anyone cares about those anyway :).
Signed-off-by: Lyude Paul <lyude@redhat.com>
Fixes: d3b2f0f792 ("drm/nouveau/kms/nv50-: Report max cursor size to userspace")
Cc: <stable@vger.kernel.org> # v5.11+
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
In preparation for Xe HP multi-tile architecture with multiple memory
regions, we need to be able differentiate multiple instances of device
local-memory.
Note that the region name is just to give it a human friendly
identifier, instead of using class/instance which also uniquely
identifies the region. So far the region name is only for our own
internal debugging in the kernel(like in the selftests), or debugfs
which prints the list of regions, including the regions name.
v2: add commentary for our current region name use
Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20210203171231.551338-1-matthew.auld@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In defer_request() we start with the request we just unsubmitted (that
should be the active request on the gpu) and then defer all of its
waiters. No waiter should be ahead of the active request, so none should
be marked as active. That assert failed.
Of particular note this machine was undergoing persistent GPU resets due
to underlying HW issues, so that may be a clue. A request is also marked
as active when it is retired, regardless of current queue status, and so
this assertion failure may be a result of the queue being completed by
the reset and then subsequently processed by the tasklet.
We can filter out retired requests here by doing the assertion check
after the is-ready check (active is a subset of being ready).
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2978
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210125140136.10494-2-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
