All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>
To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Cc: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>,
	matthew.auld@intel.com
Subject: [PATCH v3 0/7] drm/i915: Asynchronous vma unbinding
Date: Fri, 17 Dec 2021 15:52:21 +0100	[thread overview]
Message-ID: <20211217145228.10987-1-thomas.hellstrom@linux.intel.com> (raw)

This patch series introduces infrastructure for asynchronous vma
unbinding. The single enabled use-case is initially at buffer object
migration where we otherwise sync when unbinding vmas before migration.
This in theory allows us to pipeline any number of migrations, but in
practice the number is restricted by a sync wait when filling the
migration context ring. We might want to look at that moving forward if
needed.

The other main use-case is to be able to pipeline vma evictions, for
example with softpinning where a new vma wants to reuse the vm range
of an already active vma. We can't support this just yet because we
need dma_resv locking around vma eviction for that, which is under
implementation.

Patch 1 and 2 are mainly a fix and a subsequent rearrangement of code,
Patch 3 is needed for consistent bind locking,
Patch 4 introduces vma resource first for error capture purposes.
Patch 5 changes the vm backend interface to take vma resources rather than vmas,
Patch 6 introduces the async unbinding itself, and finally
Patch 7 realizes we have duplicated functionality and removes the vma snapshots.

v2:
-- Some kernel test robot reports addressed.
-- kmem cache for vma resources, See patch 6
-- Various fixes all over the place. See separate commit messages.
v3:
-- Re-add a missing i915_vma_resource_put()
-- Remove a stray debug printout

Thomas Hellström (7):
  drm/i915: Avoid using the i915_fence_array when collecting
    dependencies
  drm/i915: Break out the i915_deps utility
  drm/i915: Require the vm mutex for i915_vma_bind()
  drm/i915: Initial introduction of vma resources
  drm/i915: Use the vma resource as argument for gtt binding / unbinding
  drm/i915: Use vma resources for async unbinding
  drm/i915: Use struct vma_resource instead of struct vma_snapshot

 drivers/gpu/drm/i915/Makefile                 |   3 +-
 drivers/gpu/drm/i915/display/intel_dpt.c      |  27 +-
 .../gpu/drm/i915/gem/i915_gem_execbuffer.c    |  67 ++-
 .../gpu/drm/i915/gem/i915_gem_object_types.h  |  27 +-
 drivers/gpu/drm/i915/gem/i915_gem_ttm_move.c  | 304 ++------------
 .../gpu/drm/i915/gem/selftests/huge_pages.c   |  37 +-
 drivers/gpu/drm/i915/gt/gen6_ppgtt.c          |  19 +-
 drivers/gpu/drm/i915/gt/gen8_ppgtt.c          |  37 +-
 drivers/gpu/drm/i915/gt/intel_engine_cs.c     |   9 +-
 drivers/gpu/drm/i915/gt/intel_ggtt.c          |  72 ++--
 drivers/gpu/drm/i915/gt/intel_gtt.c           |   4 +
 drivers/gpu/drm/i915/gt/intel_gtt.h           |  18 +-
 drivers/gpu/drm/i915/gt/intel_migrate.c       |  24 +-
 drivers/gpu/drm/i915/gt/intel_migrate.h       |   9 +-
 drivers/gpu/drm/i915/gt/intel_ppgtt.c         |  22 +-
 drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c      |  13 +-
 drivers/gpu/drm/i915/gt/uc/intel_uc_fw.h      |   2 +-
 drivers/gpu/drm/i915/i915_debugfs.c           |   3 +-
 drivers/gpu/drm/i915/i915_deps.c              | 244 +++++++++++
 drivers/gpu/drm/i915/i915_deps.h              |  46 +++
 drivers/gpu/drm/i915/i915_drv.h               |   1 +
 drivers/gpu/drm/i915/i915_gem.c               |   3 +
 drivers/gpu/drm/i915/i915_gpu_error.c         |  87 ++--
 drivers/gpu/drm/i915/i915_module.c            |   3 +
 drivers/gpu/drm/i915/i915_request.c           |  34 +-
 drivers/gpu/drm/i915/i915_request.h           |   8 +-
 drivers/gpu/drm/i915/i915_vma.c               | 214 +++++++++-
 drivers/gpu/drm/i915/i915_vma.h               |  33 +-
 drivers/gpu/drm/i915/i915_vma_resource.c      | 387 ++++++++++++++++++
 drivers/gpu/drm/i915/i915_vma_resource.h      | 232 +++++++++++
 drivers/gpu/drm/i915/i915_vma_snapshot.c      | 134 ------
 drivers/gpu/drm/i915/i915_vma_snapshot.h      | 112 -----
 drivers/gpu/drm/i915/i915_vma_types.h         |   5 +
 drivers/gpu/drm/i915/selftests/i915_gem_gtt.c | 159 ++++---
 drivers/gpu/drm/i915/selftests/mock_gtt.c     |  12 +-
 35 files changed, 1571 insertions(+), 840 deletions(-)
 create mode 100644 drivers/gpu/drm/i915/i915_deps.c
 create mode 100644 drivers/gpu/drm/i915/i915_deps.h
 create mode 100644 drivers/gpu/drm/i915/i915_vma_resource.c
 create mode 100644 drivers/gpu/drm/i915/i915_vma_resource.h
 delete mode 100644 drivers/gpu/drm/i915/i915_vma_snapshot.c
 delete mode 100644 drivers/gpu/drm/i915/i915_vma_snapshot.h

-- 
2.31.1


WARNING: multiple messages have this Message-ID (diff)
From: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>
To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Cc: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>,
	matthew.auld@intel.com
Subject: [Intel-gfx] [PATCH v3 0/7] drm/i915: Asynchronous vma unbinding
Date: Fri, 17 Dec 2021 15:52:21 +0100	[thread overview]
Message-ID: <20211217145228.10987-1-thomas.hellstrom@linux.intel.com> (raw)

This patch series introduces infrastructure for asynchronous vma
unbinding. The single enabled use-case is initially at buffer object
migration where we otherwise sync when unbinding vmas before migration.
This in theory allows us to pipeline any number of migrations, but in
practice the number is restricted by a sync wait when filling the
migration context ring. We might want to look at that moving forward if
needed.

The other main use-case is to be able to pipeline vma evictions, for
example with softpinning where a new vma wants to reuse the vm range
of an already active vma. We can't support this just yet because we
need dma_resv locking around vma eviction for that, which is under
implementation.

Patch 1 and 2 are mainly a fix and a subsequent rearrangement of code,
Patch 3 is needed for consistent bind locking,
Patch 4 introduces vma resource first for error capture purposes.
Patch 5 changes the vm backend interface to take vma resources rather than vmas,
Patch 6 introduces the async unbinding itself, and finally
Patch 7 realizes we have duplicated functionality and removes the vma snapshots.

v2:
-- Some kernel test robot reports addressed.
-- kmem cache for vma resources, See patch 6
-- Various fixes all over the place. See separate commit messages.
v3:
-- Re-add a missing i915_vma_resource_put()
-- Remove a stray debug printout

Thomas Hellström (7):
  drm/i915: Avoid using the i915_fence_array when collecting
    dependencies
  drm/i915: Break out the i915_deps utility
  drm/i915: Require the vm mutex for i915_vma_bind()
  drm/i915: Initial introduction of vma resources
  drm/i915: Use the vma resource as argument for gtt binding / unbinding
  drm/i915: Use vma resources for async unbinding
  drm/i915: Use struct vma_resource instead of struct vma_snapshot

 drivers/gpu/drm/i915/Makefile                 |   3 +-
 drivers/gpu/drm/i915/display/intel_dpt.c      |  27 +-
 .../gpu/drm/i915/gem/i915_gem_execbuffer.c    |  67 ++-
 .../gpu/drm/i915/gem/i915_gem_object_types.h  |  27 +-
 drivers/gpu/drm/i915/gem/i915_gem_ttm_move.c  | 304 ++------------
 .../gpu/drm/i915/gem/selftests/huge_pages.c   |  37 +-
 drivers/gpu/drm/i915/gt/gen6_ppgtt.c          |  19 +-
 drivers/gpu/drm/i915/gt/gen8_ppgtt.c          |  37 +-
 drivers/gpu/drm/i915/gt/intel_engine_cs.c     |   9 +-
 drivers/gpu/drm/i915/gt/intel_ggtt.c          |  72 ++--
 drivers/gpu/drm/i915/gt/intel_gtt.c           |   4 +
 drivers/gpu/drm/i915/gt/intel_gtt.h           |  18 +-
 drivers/gpu/drm/i915/gt/intel_migrate.c       |  24 +-
 drivers/gpu/drm/i915/gt/intel_migrate.h       |   9 +-
 drivers/gpu/drm/i915/gt/intel_ppgtt.c         |  22 +-
 drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c      |  13 +-
 drivers/gpu/drm/i915/gt/uc/intel_uc_fw.h      |   2 +-
 drivers/gpu/drm/i915/i915_debugfs.c           |   3 +-
 drivers/gpu/drm/i915/i915_deps.c              | 244 +++++++++++
 drivers/gpu/drm/i915/i915_deps.h              |  46 +++
 drivers/gpu/drm/i915/i915_drv.h               |   1 +
 drivers/gpu/drm/i915/i915_gem.c               |   3 +
 drivers/gpu/drm/i915/i915_gpu_error.c         |  87 ++--
 drivers/gpu/drm/i915/i915_module.c            |   3 +
 drivers/gpu/drm/i915/i915_request.c           |  34 +-
 drivers/gpu/drm/i915/i915_request.h           |   8 +-
 drivers/gpu/drm/i915/i915_vma.c               | 214 +++++++++-
 drivers/gpu/drm/i915/i915_vma.h               |  33 +-
 drivers/gpu/drm/i915/i915_vma_resource.c      | 387 ++++++++++++++++++
 drivers/gpu/drm/i915/i915_vma_resource.h      | 232 +++++++++++
 drivers/gpu/drm/i915/i915_vma_snapshot.c      | 134 ------
 drivers/gpu/drm/i915/i915_vma_snapshot.h      | 112 -----
 drivers/gpu/drm/i915/i915_vma_types.h         |   5 +
 drivers/gpu/drm/i915/selftests/i915_gem_gtt.c | 159 ++++---
 drivers/gpu/drm/i915/selftests/mock_gtt.c     |  12 +-
 35 files changed, 1571 insertions(+), 840 deletions(-)
 create mode 100644 drivers/gpu/drm/i915/i915_deps.c
 create mode 100644 drivers/gpu/drm/i915/i915_deps.h
 create mode 100644 drivers/gpu/drm/i915/i915_vma_resource.c
 create mode 100644 drivers/gpu/drm/i915/i915_vma_resource.h
 delete mode 100644 drivers/gpu/drm/i915/i915_vma_snapshot.c
 delete mode 100644 drivers/gpu/drm/i915/i915_vma_snapshot.h

-- 
2.31.1


             reply	other threads:[~2021-12-17 14:52 UTC|newest]

Thread overview: 32+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-17 14:52 Thomas Hellström [this message]
2021-12-17 14:52 ` [Intel-gfx] [PATCH v3 0/7] drm/i915: Asynchronous vma unbinding Thomas Hellström
2021-12-17 14:52 ` [PATCH v3 1/7] drm/i915: Avoid using the i915_fence_array when collecting dependencies Thomas Hellström
2021-12-17 14:52   ` [Intel-gfx] " Thomas Hellström
2021-12-21 11:11   ` Matthew Auld
2021-12-21 11:11     ` [Intel-gfx] " Matthew Auld
2021-12-17 14:52 ` [PATCH v3 2/7] drm/i915: Break out the i915_deps utility Thomas Hellström
2021-12-17 14:52   ` [Intel-gfx] " Thomas Hellström
2021-12-21 11:13   ` Matthew Auld
2021-12-21 11:13     ` [Intel-gfx] " Matthew Auld
2021-12-17 14:52 ` [PATCH v3 3/7] drm/i915: Require the vm mutex for i915_vma_bind() Thomas Hellström
2021-12-17 14:52   ` [Intel-gfx] " Thomas Hellström
2021-12-21 11:32   ` Matthew Auld
2021-12-21 11:32     ` [Intel-gfx] " Matthew Auld
2021-12-17 14:52 ` [PATCH v3 4/7] drm/i915: Initial introduction of vma resources Thomas Hellström
2021-12-17 14:52   ` [Intel-gfx] " Thomas Hellström
2021-12-17 14:52 ` [PATCH v3 5/7] drm/i915: Use the vma resource as argument for gtt binding / unbinding Thomas Hellström
2021-12-17 14:52   ` [Intel-gfx] " Thomas Hellström
2021-12-17 14:52 ` [PATCH v3 6/7] drm/i915: Use vma resources for async unbinding Thomas Hellström
2021-12-17 14:52   ` [Intel-gfx] " Thomas Hellström
2021-12-21 14:02   ` Matthew Auld
2021-12-21 14:02     ` [Intel-gfx] " Matthew Auld
2021-12-21 16:07     ` Thomas Hellström
2021-12-21 16:07       ` [Intel-gfx] " Thomas Hellström
2021-12-21 17:08       ` Matthew Auld
2021-12-21 17:08         ` [Intel-gfx] " Matthew Auld
2021-12-17 14:52 ` [PATCH v3 7/7] drm/i915: Use struct vma_resource instead of struct vma_snapshot Thomas Hellström
2021-12-17 14:52   ` [Intel-gfx] " Thomas Hellström
2021-12-17 15:11 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: Asynchronous vma unbinding (rev3) Patchwork
2021-12-17 15:12 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2021-12-17 15:40 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2021-12-17 17:56 ` [Intel-gfx] ✓ Fi.CI.IGT: " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20211217145228.10987-1-thomas.hellstrom@linux.intel.com \
    --to=thomas.hellstrom@linux.intel.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=matthew.auld@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.