Use goto to handle the error path to avoid duplicating the same code. In
the error path intel_dig_port is the last one to be released as it was
the first one to be allocated and ideally the error path should be the
reverse of the execution path.
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
There is a typo in the function i915_handle_error()
kernel-doc and the word register is spelled wrongly.
Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Add the dev parameter for the functions i915_enable_asle_pipestat() and
i915_reset_and_wakeup() to the kernel-doc to fix the following warnings:
.//drivers/gpu/drm/i915/i915_irq.c:586: warning: No description found for parameter 'dev'
.//drivers/gpu/drm/i915/i915_irq.c:2400: warning: No description found for parameter 'dev'
Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
intel_rcs_ctx_init() emits all workaround register writes on the list
to the ring, in addition to calling i915_gem_render_state_init(). The
workaround list is currently empty on Gen6-7 so this shouldn't cause
any functional changes.
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
It's not an error for the workaround list to be empty if no
workarounds are needed. This will avoid spamming the logs
unnecessarily on Gen6 after the workaround list is hooked up on
pre-Gen8 hardware by the following commits.
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Passing cliprects into the kernel for it to re-execute the batch buffer
with different CMD_DRAWRECT died out long ago. As DRI1 support has been
removed from the kernel, we can now simply reject any execbuf trying to
use this "feature".
To keep Daniel happy with the prospect of being able to reuse these
fields in the next decade, continue to ensure that current userspace is
not passing garbage in through the dead fields.
v2: Fix the cliprects_ptr check
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Dave Gordon <david.s.gordon@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Exclude active GPU pages from the purview of the background shrinker
(kswapd), as these cause uncontrollable GPU stalls. Given that the
shrinker is rerun until the freelists are satisfied, we should have
opportunity in subsequent passes to recover the pages once idle. If the
machine does run out of memory entirely, we have the forced idling in the
oom-notifier as a means of releasing all the pages we can before an oom
is prematurely executed.
Note that this relies upon an up-front retire_requests to keep the
inactive list in shape, which was added in a previous patch, mostly as
execlist ctx pinning band-aids.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
[danvet: Add note about retire_requests.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
With UMS gone, we no longer use it during suspend. And with the last
user removed from the shrinker, we can remove the dead code.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We can forgo an evict-everything here as the shrinker operation itself
will unbind any vma as required. If we explicitly idle the GPU through a
switch to the default context, we not only create a request in an
illegal context (e.g. whilst shrinking during execbuf with a request
already allocated), but switching to the default context will not free
up the memory backing the active contexts - unless in the unlikely
situation that context had already been closed (and just kept arrive by
being the current context). The saving is near zero and the danger real.
To compensate for the loss of the forced retire, add a couple of
retire-requests to i915_gem_shirnk() - this should help free up any
transitive cache from the requests.
Note that the second retire_requests is for the benefit of the
hand-rolled execlist ctx active tracking: We need to manually kick
requests to get those unpinned again. Once that's fixed we can try to
remove this again.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: Add summary of why we need a pile of retire_requests.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Often it is very useful to know why we suddenly purge vast tracts of
memory and surprisingly up until now we didn't even have a tracepoint
for when we shrink our memory.
Note that there are slab_start/end tracepoints already, but those
don't cover the internal recursion when we directly call into our
shrinker code. Hence a separate tracepoint seems justified. Also note
that we don't really need a separate tracepoint for the actual amount
of pages freed since we already have an unbind tracpoint for that.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: Add a note that there's also slab_start/end and why they're
insufficient.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Add the item of i915_component.h in DocBook and add the DOC for
i915_component.h. Explain the struct i915_audio_component_ops and
struct i915_audio_component_audio_ops usage.
Signed-off-by: Libin Yang <libin.yang@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Pull in the i915/hda changes for N/CTS setting so I can apply the
follow-up documentation work for drm/i915.
Some conflicts because ofc we had to rework i915 while that N/CTS work
was going on. But not more than adjacent changes really.
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
As the shrinker_control now passes us unsigned long targets, update our
shrinker functions to match.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
I've botched this, so let's fix it.
Botched in
commit eb0b44adc0
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Wed Mar 18 14:47:59 2015 +0100
drm/i915: kerneldoc for i915_gem_shrinker.c
v2: Be a good citizen^Wmaintainer and add the proper commit citation.
Noticed by Jani.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Whilst discussing possible ways to trigger an invalidate_range on a
userptr with an aliased GGTT mmapping (and so cause a struct_mutex
deadlock), the conclusion is that we can, and we must, prevent any
possible deadlock by avoiding taking the mutex at all during
invalidate_range. This has numerous advantages all of which stem from
avoid the sleeping function from inside the unknown context. In
particular, it simplifies the invalidate_range because we no longer
have to juggle the spinlock/mutex and can just hold the spinlock
for the entire walk. To compensate, we have to make get_pages a bit more
complicated in order to serialise with a pending cancel_userptr worker.
As we hold the struct_mutex, we have no choice but to return EAGAIN and
hope that the worker is then flushed before we retry after reacquiring
the struct_mutex.
The important caveat is that the invalidate_range itself is no longer
synchronous. There exists a small but definite period in time in which
the old PTE's page remain accessible via the GPU. Note however that the
physical pages themselves are not invalidated by the mmu_notifier, just
the CPU view of the address space. The impact should be limited to a
delay in pages being flushed, rather than a possibility of writing to
the wrong pages. The only race condition that this worsens is remapping
an userptr active on the GPU where fresh work may still reference the
old pages due to struct_mutex contention. Given that userspace is racing
with the GPU, it is fair to say that the results are undefined.
v2: Only queue (and importantly only take one refcnt) the worker once.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Michał Winiarski found a really evil way to trigger a struct_mutex
deadlock with userptr. He found that if he allocated a userptr bo and
then GTT mmaped another bo, or even itself, at the same address as the
userptr using MAP_FIXED, he could then cause a deadlock any time we then
had to invalidate the GTT mmappings (so at will). Tvrtko then found by
repeatedly allocating GTT mmappings he could alias with an old userptr
mmap and also trigger the deadlock.
To counter act the deadlock, we make the observation that we only need
to take the struct_mutex if the object has any pages to revoke, and that
before userspace can alias with the userptr address space, it must have
invalidated the userptr->pages. Thus if we can check for those pages
outside of the struct_mutex, we can avoid the deadlock. To do so we
introduce a separate flag for userptr objects that we can inspect from
the mmu-notifier underneath its spinlock.
The patch makes one eye-catching change. That is the removal serial=0
after detecting a to-be-freed object inside the invalidate walker. I
felt setting serial=0 was a questionable pessimisation: it denies us the
chance to reuse the current iterator for the next loop (before it is
freed) and being explicit makes the reader question the validity of the
locking (since the object-free race could occur elsewhere). The
serialisation of the iterator is through the spinlock, if the object is
freed before the next loop then the notifier.serial will be incremented
and we start the walk from the beginning as we detect the invalid cache.
To try and tame the error paths and interactions with the userptr->active
flag, we have to do a fair amount of rearranging of get_pages_userptr().
v2: Grammar fixes
v3: Reorder set-active so that it is only set when obj->pages is set
(and so needs cancellation). Only the order of setting obj->pages and
the active-flag is crucial. Calling gup after invalidate-range begin
means the userptr sees the new set of backing storage (and so will not
need to invalidate its new pages), but we have to be careful not to set
the active-flag prior to successfully establishing obj->pages.
v4: Take the active->flag early so we know in the mmu-notifier when we
have to cancel a pending gup-worker.
v5: Rearrange the error path so that is not so convoluted
v6: Set pinned to 0 when negative before calling release_pages()
Reported-by: Michał Winiarski <michal.winiarski@intel.com>
Testcase: igt/gem_userptr_blits/map-fixed*
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: stable@vger.kernel.org
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The userptr worker allows for a slight race condition where upon there
may two or more threads calling get_user_pages for the same object. When
we have the array of pages, then we serialise the update of the object.
However, the worker should only overwrite the obj->userptr.work pointer
if and only if it is the active one. Currently we clear it for a
secondary worker with the effect that we may rarely force a second
lookup.
v2: Rebase and rename a variable to avoid 80cols
v3: Mention v2
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We tried to fix this in commit fdc454c148 ("drm/i915: Prevent out of
range pt in gen6_for_each_pde").
But the static analyzer still complains that, just before we break due
to "iter < I915_PDES", we do "pt = (pd)->page_table[iter]" with an
iter value that is bigger than I915_PDES. Of course, this isn't really
a problem since no one uses pt outside the macro. Still, every single
new usage of the macro will create a new issue for us to mark as a
false positive.
Also, Paulo re-started the discussion a while ago [1], but didn't end up
implemented.
In order to "solve" this "problem", this patch takes the ideas from
Chris and Dave, but that check would change the desired behavior of the
code, because the object (for example pdp->page_directory[iter]) can be
null during init/alloc, and C would take this as false, breaking the for
loop immediately.
This has been already verified with "static analysis tools".
[1]http://lists.freedesktop.org/archives/intel-gfx/2015-June/068548.html
v2: Make it a single statement, while preventing the common subexpression
elimination (Chris)
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Dave Gordon <david.s.gordon@intel.com>
Signed-off-by: Michel Thierry <michel.thierry@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Prevent leaking VMAs and PPGTT VMs when objects are imported
via flink.
Scenario is that any VMAs created by the importer will be left
dangling after the importer exits, or destroys the PPGTT context
with which they are associated.
This is caused by object destruction not running when the
importer closes the buffer object handle due the reference held
by the exporter. This also leaks the VM since the VMA has a
reference on it.
In practice these leaks can be observed by stopping and starting
the X server on a kernel with fbcon compiled in. Every time
X server exits another VMA will be leaked against the fbcon's
frame buffer object.
Also on systems where flink buffer sharing is used extensively,
like Android, this leak has even more serious consequences.
This version is takes a general approach from the earlier work
by Rafael Barbalho (drm/i915: Clean-up PPGTT on context
destruction) and tries to incorporate the subsequent discussion
between Chris Wilson and Daniel Vetter.
v2:
Removed immediate cleanup on object retire - it was causing a
recursive VMA unbind via i915_gem_object_wait_rendering. And
it is in fact not even needed since by definition context
cleanup worker runs only after the last context reference has
been dropped, hence all VMAs against the VM belonging to the
context are already on the inactive list.
v3:
Previous version could deadlock since VMA unbind waits on any
rendering on an object to complete. Objects can be busy in a
different VM which would mean that the cleanup loop would do
the wait with the struct mutex held.
This is an even simpler approach where we just unbind VMAs
without waiting since we know all VMAs belonging to this VM
are idle, and there is nothing in flight, at the point
context destructor runs.
v4:
Double underscore prefix for __915_vma_unbind_no_wait and a
commit message typo fix. (Michel Thierry)
Note that this is just a partial/interim fix since we have a bit a
fundamental issue with cleaning up, e.g.
https://bugs.freedesktop.org/show_bug.cgi?id=87729
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Testcase: igt/gem_ppgtt.c/flink-and-exit-vma-leak
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Rafael Barbalho <rafael.barbalho@intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
[danvet: Add a note that this isn't everything.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Sudip Mukherjee
Javier Martinez Canillas
Francisco Jerez
Daniel Vetter
Jani Nikula
Akash Goel
Chris Wilson
Libin Yang
Michel Thierry
Tvrtko Ursulin
kbuild test robot