[PATCH 00/27] drm/i915/gem: Rework i915_gem_object locking to use ww locks.

[Maarten Lankhorst]

We want to get rid of obj->mm.lock, but for this we need to start using
the ww locking everywhere.

First revert a bunch fo patches that conflict with our goal, async
relocations should not be a goal, since the only user is media, which
can be converted to not require relocations at all.

I think we want to get rid of gpu relocations altogether, as it
complicates the relocation code a lot wrt extra object creation and
tracking.

This also changes the ordering around ctx->timeline and ww, ww becomes
the outermost lock, while the timeline lock becomes the inner lock.
This means that we will get lockdep splats of those between
  "drm/i915: Use per object locking in execbuf, v12." and
  "drm/i915: Move i915_vma_lock in the selftests to avoid lock inversion, v3."

There is 1 abuser left of lockdep, which is userptr. I changed it to
be similar to amdgpu userptr code, but this is untested so I leave it
in for CI results mostly.

When userptr is fixed, we are very close to the goal of removing
obj->mm.lock entirely, this should be a near future goal of this
conversion.

Maarten Lankhorst (27):
  Revert "drm/i915/gem: Async GPU relocations only"
  drm/i915: Revert relocation chaining commits.
  Revert "drm/i915/gem: Drop relocation slowpath".
  Revert "drm/i915/gem: Split eb_vma into its own allocation"
  drm/i915: Add an implementation for i915_gem_ww_ctx locking, v2.
  drm/i915: Remove locking from i915_gem_object_prepare_read/write
  drm/i915: Parse command buffer earlier in eb_relocate(slow)
  drm/i915: Use per object locking in execbuf, v12.
  drm/i915: Use ww locking in intel_renderstate.
  drm/i915: Add ww context handling to context_barrier_task
  drm/i915: Nuke arguments to eb_pin_engine
  drm/i915: Pin engine before pinning all objects, v5.
  drm/i915: Rework intel_context pinning to do everything outside of
    pin_mutex
  drm/i915: Make sure execbuffer always passes ww state to i915_vma_pin.
  drm/i915: Convert i915_gem_object/client_blt.c to use ww locking as
    well, v2.
  drm/i915: Kill last user of intel_context_create_request outside of
    selftests
  drm/i915: Convert i915_perf to ww locking as well
  drm/i915: Dirty hack to fix selftests locking inversion
  drm/i915/selftests: Fix locking inversion in lrc selftest.
  drm/i915: Use ww pinning for intel_context_create_request()
  drm/i915: Move i915_vma_lock in the selftests to avoid lock inversion,
    v3.
  drm/i915: Add ww locking to vm_fault_gtt
  drm/i915: Add ww locking to pin_to_display_plane
  drm/i915: Ensure we hold the object mutex in pin correctly.
  drm/i915: Add gem object locking to madvise.
  drm/i915: Convert i915_gem_object_attach_phys() to ww locking
  drm/i915: Fix userptr so we do not have to worry about obj->mm.lock

 drivers/gpu/drm/i915/display/intel_display.c  |    8 +-
 .../gpu/drm/i915/gem/i915_gem_client_blt.c    |   78 +-
 drivers/gpu/drm/i915/gem/i915_gem_context.c   |   55 +-
 drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c    |    4 +-
 drivers/gpu/drm/i915/gem/i915_gem_domain.c    |   89 +-
 .../gpu/drm/i915/gem/i915_gem_execbuffer.c    | 1594 +++++++++++------
 drivers/gpu/drm/i915/gem/i915_gem_mman.c      |   51 +-
 drivers/gpu/drm/i915/gem/i915_gem_object.h    |   51 +-
 .../gpu/drm/i915/gem/i915_gem_object_blt.c    |  152 +-
 .../gpu/drm/i915/gem/i915_gem_object_blt.h    |    3 +
 .../gpu/drm/i915/gem/i915_gem_object_types.h  |   16 +-
 drivers/gpu/drm/i915/gem/i915_gem_phys.c      |   35 +-
 drivers/gpu/drm/i915/gem/i915_gem_pm.c        |    2 +-
 drivers/gpu/drm/i915/gem/i915_gem_tiling.c    |    2 +-
 drivers/gpu/drm/i915/gem/i915_gem_userptr.c   |  691 ++-----
 .../gpu/drm/i915/gem/selftests/huge_pages.c   |    7 +-
 .../i915/gem/selftests/i915_gem_client_blt.c  |    2 +-
 .../i915/gem/selftests/i915_gem_coherency.c   |   50 +-
 .../drm/i915/gem/selftests/i915_gem_context.c |  142 +-
 .../i915/gem/selftests/i915_gem_execbuffer.c  |   60 +-
 .../drm/i915/gem/selftests/i915_gem_mman.c    |   45 +-
 .../drm/i915/gem/selftests/i915_gem_phys.c    |    2 +-
 drivers/gpu/drm/i915/gt/gen6_ppgtt.c          |    4 +-
 drivers/gpu/drm/i915/gt/gen6_ppgtt.h          |    4 +-
 drivers/gpu/drm/i915/gt/intel_context.c       |  309 ++--
 drivers/gpu/drm/i915/gt/intel_context.h       |   13 +
 drivers/gpu/drm/i915/gt/intel_context_types.h |    5 +-
 drivers/gpu/drm/i915/gt/intel_engine_cs.c     |    2 +-
 drivers/gpu/drm/i915/gt/intel_gt.c            |   23 +-
 drivers/gpu/drm/i915/gt/intel_lrc.c           |   37 +-
 drivers/gpu/drm/i915/gt/intel_renderstate.c   |   75 +-
 drivers/gpu/drm/i915/gt/intel_renderstate.h   |    9 +-
 drivers/gpu/drm/i915/gt/intel_ring.c          |   10 +-
 drivers/gpu/drm/i915/gt/intel_ring.h          |    3 +-
 .../gpu/drm/i915/gt/intel_ring_submission.c   |   20 +-
 drivers/gpu/drm/i915/gt/intel_timeline.c      |   12 +-
 drivers/gpu/drm/i915/gt/intel_timeline.h      |    3 +-
 drivers/gpu/drm/i915/gt/intel_workarounds.c   |   43 +-
 drivers/gpu/drm/i915/gt/mock_engine.c         |   14 +-
 drivers/gpu/drm/i915/gt/selftest_lrc.c        |   17 +-
 drivers/gpu/drm/i915/gt/selftest_rps.c        |   30 +-
 drivers/gpu/drm/i915/gt/selftest_timeline.c   |    4 +-
 .../gpu/drm/i915/gt/selftest_workarounds.c    |    2 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc.c        |    2 +-
 drivers/gpu/drm/i915/gvt/cmd_parser.c         |    3 +-
 drivers/gpu/drm/i915/i915_drv.h               |   22 +-
 drivers/gpu/drm/i915/i915_gem.c               |   97 +-
 drivers/gpu/drm/i915/i915_gem.h               |   12 +
 drivers/gpu/drm/i915/i915_perf.c              |   57 +-
 drivers/gpu/drm/i915/i915_vma.c               |   20 +-
 drivers/gpu/drm/i915/i915_vma.h               |   16 +-
 drivers/gpu/drm/i915/selftests/i915_gem.c     |   41 +
 drivers/gpu/drm/i915/selftests/i915_request.c |   18 +-
 drivers/gpu/drm/i915/selftests/i915_vma.c     |    2 +-
 .../drm/i915/selftests/intel_memory_region.c  |    2 +-
 55 files changed, 2469 insertions(+), 1601 deletions(-)


base-commit: 943d034c433e5be93076cf51fd8ea5b4d7644e8b
-- 
2.28.0.rc2