On Fri, Jun 13, 2014 at 04:37:31PM +0100, oscar.mateo@xxxxxxxxx wrote:
> From: Oscar Mateo <oscar.mateo@xxxxxxxxx>
>
> The backing objects for contexts created via open fd are actually
> empty until the user starts sending execbuffers to them. We do this
> because, at create time, we really don't know which engine is going
> to be used with the context later on.
>
> v2: As context created via ioctl can only be used with the render ring,
> we have enough information to allocate & populate them right away.
Not sure this is a good choice to special-case the ioctl. We might want
to allow contexts also on non-render rings, and it introduces a special
case for not much gain. Since we must have the deferred alloca anyway lets
use that for all cases. And if there's a (performance) issue with it we
need to address that. One true (code) path to rule them all.
-Daniel
>
> Signed-off-by: Oscar Mateo <oscar.mateo@xxxxxxxxx>
> ---
> drivers/gpu/drm/i915/i915_gem_context.c | 28 ++++++++++++++++++++++++++--
> 1 file changed, 26 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
> index 1fb4592..70bf6d0 100644
> --- a/drivers/gpu/drm/i915/i915_gem_context.c
> +++ b/drivers/gpu/drm/i915/i915_gem_context.c
> @@ -795,6 +795,7 @@ static bool contexts_enabled(struct drm_device *dev)
> int i915_gem_context_create_ioctl(struct drm_device *dev, void *data,
> struct drm_file *file)
> {
> + struct drm_i915_private *dev_priv = dev->dev_private;
> struct drm_i915_gem_context_create *args = data;
> struct drm_i915_file_private *file_priv = file->driver_priv;
> struct intel_context *ctx;
> @@ -808,9 +809,23 @@ int i915_gem_context_create_ioctl(struct drm_device *dev, void *data,
> return ret;
>
> ctx = i915_gem_create_context(dev, file_priv, USES_FULL_PPGTT(dev));
> - mutex_unlock(&dev->struct_mutex);
> - if (IS_ERR(ctx))
> + if (IS_ERR(ctx)) {
> + mutex_unlock(&dev->struct_mutex);
> return PTR_ERR(ctx);
> + }
> +
> + if (dev_priv->lrc_enabled) {
> + /* NB: We know this context will only be used with the render ring
> + * (as we enforce it) so we can allocate & populate it already */
> + int ret = intel_lr_context_deferred_create(ctx, &dev_priv->ring[RCS]);
> + if (ret) {
> + mutex_unlock(&dev->struct_mutex);
> + DRM_DEBUG_DRIVER("Could not create LRC: %d\n", ret);
> + return ret;
> + }
> + }
> +
> + mutex_unlock(&dev->struct_mutex);
>
> args->ctx_id = ctx->id;
> DRM_DEBUG_DRIVER("HW context %d created\n", args->ctx_id);
> @@ -851,6 +866,7 @@ struct intel_context *
> i915_gem_context_validate(struct drm_device *dev, struct drm_file *file,
> struct intel_engine_cs *ring, const u32 ctx_id)
> {
> + struct drm_i915_private *dev_priv = dev->dev_private;
> struct intel_context *ctx = NULL;
> struct i915_ctx_hang_stats *hs;
>
> @@ -867,5 +883,13 @@ i915_gem_context_validate(struct drm_device *dev, struct drm_file *file,
> return ERR_PTR(-EIO);
> }
>
> + if (dev_priv->lrc_enabled && !ctx->engine[ring->id].obj) {
> + int ret = intel_lr_context_deferred_create(ctx, ring);
> + if (ret) {
> + DRM_DEBUG("Could not create LRC %u: %d\n", ctx_id, ret);
> + return ERR_PTR(ret);
> + }
> + }
> +
> return ctx;
> }
> --
> 1.9.0
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
