[Intel-gfx] [PATCH v4 7/7] drm/i915: Tidy up execbuffer command parsing code
Chris Wilson
chris at chris-wilson.co.uk
Wed Nov 12 10:37:29 CET 2014
On Fri, Nov 07, 2014 at 02:22:07PM -0800, bradley.d.volkin at intel.com wrote:
> From: Brad Volkin <bradley.d.volkin at intel.com>
>
> Move it to a separate function since the main do_execbuffer function
> already has so much going on.
>
> Signed-off-by: Brad Volkin <bradley.d.volkin at intel.com>
> ---
> drivers/gpu/drm/i915/i915_gem_execbuffer.c | 136 +++++++++++++++++------------
> 1 file changed, 79 insertions(+), 57 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> index a271bc0..58f0a6c 100644
> --- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> +++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> @@ -1026,6 +1026,75 @@ i915_reset_gen7_sol_offsets(struct drm_device *dev,
> return 0;
> }
>
> +static struct drm_i915_gem_object*
> +i915_gem_execbuffer_parse(struct intel_engine_cs *ring,
> + struct drm_i915_gem_exec_object2 *shadow_exec_entry,
> + struct eb_vmas *eb,
> + struct drm_i915_gem_object *batch_obj,
> + u32 batch_start_offset,
> + u32 batch_len,
> + bool is_master,
> + u32 *flags)
> +{
> + struct drm_i915_private *dev_priv = to_i915(batch_obj->base.dev);
> + struct drm_i915_gem_object *shadow_batch_obj;
> + int ret;
> +
> + shadow_batch_obj = i915_gem_batch_pool_get(&dev_priv->mm.batch_pool,
> + batch_obj->base.size);
> + if (IS_ERR(shadow_batch_obj))
> + return shadow_batch_obj;
> +
> + shadow_batch_obj->madv = I915_MADV_WILLNEED;
> +
> + ret = i915_gem_obj_ggtt_pin(shadow_batch_obj, 4096, 0);
> + if (ret)
> + goto err;
Pardon? This feels an implementation issue of i915_parse_cmds() and should
be resolved there. Presumably you are not actually reading back through
the GTT? That would be insane...
> + ret = i915_parse_cmds(ring,
> + batch_obj,
> + shadow_batch_obj,
> + batch_start_offset,
> + batch_len,
> + is_master);
> + i915_gem_object_ggtt_unpin(shadow_batch_obj);
Yet pin+unpin around the parser seems to serve no other purpose.
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
More information about the Intel-gfx
mailing list