1. 19 12月, 2016 2 次提交
    • C
      drm/i915: Simplify releasing context reference · 69df05e1
      Chris Wilson 提交于
      A few users only take the struct_mutex in order to release a reference
      to a context. We can expose a kref_put_mutex() wrapper in order to
      simplify these users, and optimise taking of the mutex to the final
      unref.
      Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
      Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
      Link: http://patchwork.freedesktop.org/patch/msgid/20161218153724.8439-4-chris@chris-wilson.co.uk
      69df05e1
    • C
      drm/i915: Unify active context tracking between legacy/execlists/guc · e8a9c58f
      Chris Wilson 提交于
      The requests conversion introduced a nasty bug where we could generate a
      new request in the middle of constructing a request if we needed to idle
      the system in order to evict space for a context. The request to idle
      would be executed (and waited upon) before the current one, creating a
      minor havoc in the seqno accounting, as we will consider the current
      request to already be completed (prior to deferred seqno assignment) but
      ring->last_retired_head would have been updated and still could allow
      us to overwrite the current request before execution.
      
      We also employed two different mechanisms to track the active context
      until it was switched out. The legacy method allowed for waiting upon an
      active context (it could forcibly evict any vma, including context's),
      but the execlists method took a step backwards by pinning the vma for
      the entire active lifespan of the context (the only way to evict was to
      idle the entire GPU, not individual contexts). However, to circumvent
      the tricky issue of locking (i.e. we cannot take struct_mutex at the
      time of i915_gem_request_submit(), where we would want to move the
      previous context onto the active tracker and unpin it), we take the
      execlists approach and keep the contexts pinned until retirement.
      The benefit of the execlists approach, more important for execlists than
      legacy, was the reduction in work in pinning the context for each
      request - as the context was kept pinned until idle, it could short
      circuit the pinning for all active contexts.
      
      We introduce new engine vfuncs to pin and unpin the context
      respectively. The context is pinned at the start of the request, and
      only unpinned when the following request is retired (this ensures that
      the context is idle and coherent in main memory before we unpin it). We
      move the engine->last_context tracking into the retirement itself
      (rather than during request submission) in order to allow the submission
      to be reordered or unwound without undue difficultly.
      
      And finally an ulterior motive for unifying context handling was to
      prepare for mock requests.
      
      v2: Rename to last_retired_context, split out legacy_context tracking
      for MI_SET_CONTEXT.
      Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
      Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
      Link: http://patchwork.freedesktop.org/patch/msgid/20161218153724.8439-3-chris@chris-wilson.co.uk
      e8a9c58f
  2. 17 12月, 2016 1 次提交
  3. 12 12月, 2016 1 次提交
  4. 09 12月, 2016 2 次提交
    • R
      drm/i915/perf: More documentation hooked to i915.rst · 16d98b31
      Robert Bragg 提交于
      This adds a 'Perf' section to i915.rst with the following sub sections:
      - Overview
      - Comparison with Core Perf
      - i915 Driver Entry Points
      - i915 Perf Stream
      - i915 Perf Observation Architecture Stream
      - All i915 Perf Internals
      
      v2:
          section headers in i915.rst (Daniel Vetter)
          missing symbol docs + other fixups (Matthew Auld)
      Signed-off-by: NRobert Bragg <robert@sixbynine.org>
      Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
      Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
      Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
      Link: http://patchwork.freedesktop.org/patch/msgid/20161207214033.3581-1-robert@sixbynine.org
      16d98b31
    • I
      drm/i915/gen9: Fix PCODE polling during CDCLK change notification · a0b8a1fe
      Imre Deak 提交于
      commit 848496e5
      Author: Ville Syrjälä <ville.syrjala@linux.intel.com>
      Date:   Wed Jul 13 16:32:03 2016 +0300
      
          drm/i915: Wait up to 3ms for the pcu to ack the cdclk change request on SKL
      
      increased the timeout to match the spec, but we still see a timeout on
      at least one SKL. A CDCLK change request following the failed one will
      succeed nevertheless.
      
      I could reproduce this problem easily by running kms_pipe_crc_basic in a
      loop. In all failure cases _wait_for() was pre-empted for >3ms and so in
      the worst case - when the pre-emption happened right after calculating
      timeout__ in _wait_for() - we called skl_cdclk_wait_for_pcu_ready() only
      once which failed and so _wait_for() timed out. As opposed to this the
      spec says to keep retrying the request for at most a 3ms period.
      
      To fix this send the first request explicitly to guarantee that there is
      3ms between the first and last request. Though this matches the spec, I
      noticed that in rare cases this can still time out if we sent only a few
      requests (in the worst case 2) _and_ PCODE is busy for some reason even
      after a previous request and a 3ms delay. To work around this retry the
      polling with pre-emption disabled to maximize the number of requests.
      Also increase the timeout to 10ms to account for interrupts that could
      reduce the number of requests. With this change I couldn't trigger
      the problem.
      
      v2:
      - Use 1ms poll period instead of 10us. (Chris)
      v3:
      - Poll with pre-emption disabled to increase the number of request
        attempts. (Ville, Chris)
      - Factor out a helper to poll, it's also needed by the next patch.
      v4:
      - Pass reply_mask, reply to skl_pcode_request(), instead of assuming the
        reply is generic. (Ville)
      v5:
      - List the request specific timeout values as code comment. (Ville)
      v6:
      - Try the poll first with preemption enabled.
      - Add code comment about first request being queued by PCODE. (Art)
      - Add timeout_base_ms argument. (Ville)
      v7:
      - Clarify code comment about first queued request. (Chris)
      
      Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
      Cc: Chris Wilson <chris@chris-wilson.co.uk>
      Cc: Art Runyan <arthur.j.runyan@intel.com>
      Cc: <stable@vger.kernel.org> # v4.2- : 3b2c1710 : drm/i915: Wait up to 3ms
      Cc: <stable@vger.kernel.org> # v4.2-
      Fixes: 5d96d8af ("drm/i915/skl: Deinit/init the display at suspend/resume")
      Reference: https://bugs.freedesktop.org/show_bug.cgi?id=97929
      Testcase: igt/kms_pipe_crc_basic/suspend-read-crc-pipe-B
      Signed-off-by: NImre Deak <imre.deak@intel.com>
      Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
      Link: http://patchwork.freedesktop.org/patch/msgid/1480955258-26311-1-git-send-email-imre.deak@intel.com
      a0b8a1fe
  5. 08 12月, 2016 3 次提交
  6. 07 12月, 2016 8 次提交
  7. 06 12月, 2016 2 次提交
    • C
      drm/i915: Fix i915_gem_evict_for_vma (soft-pinning) · 172ae5b4
      Chris Wilson 提交于
      Soft-pinning depends upon being able to check for availabilty of an
      interval and evict overlapping object from a drm_mm range manager very
      quickly. Currently it uses a linear list, and so performance is dire and
      not suitable as a general replacement. Worse, the current code will oops
      if it tries to evict an active buffer.
      
      It also helps if the routine reports the correct error codes as expected
      by its callers and emits a tracepoint upon use.
      
      For posterity since the wrong patch was pushed (i.e. that missed these
      key points and had known bugs), this is the changelog that should have
      been on commit 506a8e87 ("drm/i915: Add soft-pinning API for
      execbuffer"):
      
      Userspace can pass in an offset that it presumes the object is located
      at. The kernel will then do its utmost to fit the object into that
      location. The assumption is that userspace is handling its own object
      locations (for example along with full-ppgtt) and that the kernel will
      rarely have to make space for the user's requests.
      
      This extends the DRM_IOCTL_I915_GEM_EXECBUFFER2 to do the following:
      * if the user supplies a virtual address via the execobject->offset
        *and* sets the EXEC_OBJECT_PINNED flag in execobject->flags, then
        that object is placed at that offset in the address space selected
        by the context specifier in execbuffer.
      * the location must be aligned to the GTT page size, 4096 bytes
      * as the object is placed exactly as specified, it may be used by this
        execbuffer call without relocations pointing to it
      
      It may fail to do so if:
      * EINVAL is returned if the object does not have a 4096 byte aligned
        address
      * the object conflicts with another pinned object (either pinned by
        hardware in that address space, e.g. scanouts in the aliasing ppgtt)
        or within the same batch.
        EBUSY is returned if the location is pinned by hardware
        EINVAL is returned if the location is already in use by the batch
      * EINVAL is returned if the object conflicts with its own alignment (as meets
        the hardware requirements) or if the placement of the object does not fit
        within the address space
      
      All other execbuffer errors apply.
      
      Presence of this execbuf extension may be queried by passing
      I915_PARAM_HAS_EXEC_SOFTPIN to DRM_IOCTL_I915_GETPARAM and checking for
      a reported value of 1 (or greater).
      
      v2: Combine the hole/adjusted-hole ENOSPC checks
      v3: More color, more splitting, more blurb.
      
      Fixes: 506a8e87 ("drm/i915: Add soft-pinning API for execbuffer")
      Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
      Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
      Link: http://patchwork.freedesktop.org/patch/msgid/20161205142941.21965-2-chris@chris-wilson.co.uk
      172ae5b4
    • C
      drm/i915: Mark all non-vma being inserted into the address spaces · 85fd4f58
      Chris Wilson 提交于
      We need to distinguish between full i915_vma structs and simple
      drm_mm_nodes when considering eviction (i.e. we must be careful not to
      treat a mere drm_mm_node as a much larger i915_vma causing memory
      corruption, if we are lucky). To do this, color these not-a-vma with -1
      (I915_COLOR_UNEVICTABLE).
      
      v2...v200: New name for -1.
      Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
      Reviewed-by: NJoonas Lahtinen <joonas.lahtinen@linux.intel.com>
      Link: http://patchwork.freedesktop.org/patch/msgid/20161205142941.21965-1-chris@chris-wilson.co.uk
      85fd4f58
  8. 05 12月, 2016 3 次提交
  9. 02 12月, 2016 9 次提交
  10. 01 12月, 2016 2 次提交
  11. 26 11月, 2016 1 次提交
  12. 25 11月, 2016 2 次提交
  13. 24 11月, 2016 4 次提交