Re: [Intel-gfx] [RFC PATCH 14/42] drm/i915: introduce kernel blitter_context

[Chris Wilson]

Quoting Matthew Auld (2019-02-14 14:57:12)
> We may be without a context to perform various internal blitter
> operations, for example when performing object migration. Piggybacking
> off the kernel_context is probably a bad idea, since it has other uses.

Explain why it is a worse idea than creating yet another kernel context
to manage. The fact that we may use it for blitting makes the kernel
context even more viable for its other jobs (such as knowing we've
flushed off user contexts). Convince me that we really need more than
one internal context; and that blitting especially needs its own.
-Chris
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [RFC PATCH 14/42] drm/i915: introduce kernel blitter_context

[Matthew Auld]

We may be without a context to perform various internal blitter
operations, for example when performing object migration. Piggybacking
off the kernel_context is probably a bad idea, since it has other uses.

Signed-off-by: Matthew Auld 
Cc: Joonas Lahtinen 
Cc: Abdiel Janulgue 
---
 drivers/gpu/drm/i915/i915_drv.h  |  2 ++
 drivers/gpu/drm/i915/i915_gem_context.c  | 13 +
 drivers/gpu/drm/i915/selftests/mock_gem_device.c |  4 
 3 files changed, 19 insertions(+)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index feec3bab0f5f..5b39af57c36c 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1544,6 +1544,8 @@ struct drm_i915_private {
struct intel_engine_cs *engine[I915_NUM_ENGINES];
/* Context used internally to idle the GPU and setup initial state */
struct i915_gem_context *kernel_context;
+   /* Context used internally for various blitting operations */
+   struct i915_gem_context *blitter_context;
/* Context only to be used for injecting preemption commands */
struct i915_gem_context *preempt_context;
struct intel_engine_cs *engine_class[MAX_ENGINE_CLASS + 1]
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c 
b/drivers/gpu/drm/i915/i915_gem_context.c
index 280813a4bf82..e814030dd89a 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -583,6 +583,18 @@ int i915_gem_contexts_init(struct drm_i915_private 
*dev_priv)
GEM_BUG_ON(!atomic_read(&ctx->hw_id_pin_count));
dev_priv->kernel_context = ctx;
 
+   ctx = i915_gem_context_create_kernel(dev_priv, I915_PRIORITY_NORMAL);
+   if (IS_ERR(ctx)) {
+   DRM_ERROR("Failed to create blitter global context\n");
+   return PTR_ERR(ctx);
+   }
+
+   /*
+* XXX: one idea here is do s/kernel_context/idle_context/, then use the
+* kernel_context for doing things like blitting...
+*/
+   dev_priv->blitter_context = ctx;
+
/* highest priority; preempting task */
if (needs_preempt_context(dev_priv)) {
ctx = i915_gem_context_create_kernel(dev_priv, INT_MAX);
@@ -615,6 +627,7 @@ void i915_gem_contexts_fini(struct drm_i915_private *i915)
 
if (i915->preempt_context)
destroy_kernel_context(&i915->preempt_context);
+   destroy_kernel_context(&i915->blitter_context);
destroy_kernel_context(&i915->kernel_context);
 
/* Must free all deferred contexts (via flush_workqueue) first */
diff --git a/drivers/gpu/drm/i915/selftests/mock_gem_device.c 
b/drivers/gpu/drm/i915/selftests/mock_gem_device.c
index f8901cd12180..0b59246c99c8 100644
--- a/drivers/gpu/drm/i915/selftests/mock_gem_device.c
+++ b/drivers/gpu/drm/i915/selftests/mock_gem_device.c
@@ -248,6 +248,10 @@ struct drm_i915_private *mock_gem_device(void)
if (!i915->engine[RCS])
goto err_context;
 
+   i915->blitter_context = mock_context(i915, NULL);
+   if (!i915->blitter_context)
+   goto err_context;
+
mutex_unlock(&i915->drm.struct_mutex);
 
WARN_ON(i915_gemfs_init(i915));
-- 
2.20.1

___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx