[Intel-gfx] [RFC 03/11] drm/i915: Add reset stats entry point for per-engine reset.
Dave Gordon
david.s.gordon at intel.com
Thu Jun 11 02:14:18 PDT 2015
On 08/06/15 18:03, Tomas Elf wrote:
> In preparation for per-engine reset add way for setting context reset stats.
>
> OPEN QUESTIONS:
> 1. How do we deal with get_reset_stats and the GL robustness interface when
> introducing per-engine resets?
>
> a. Do we set context that cause per-engine resets as guilty? If so, how
> does this affect context banning?
>
> b. Do we extend the publically available reset stats to also contain
> per-engine reset statistics? If so, would this break the ABI?
>
> Signed-off-by: Tomas Elf <tomas.elf at intel.com>
> ---
> drivers/gpu/drm/i915/i915_drv.h | 2 ++
> drivers/gpu/drm/i915/i915_gem.c | 14 ++++++++++++++
> 2 files changed, 16 insertions(+)
>
> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> index 47be4a5..ab5dfdc 100644
> --- a/drivers/gpu/drm/i915/i915_drv.h
> +++ b/drivers/gpu/drm/i915/i915_drv.h
> @@ -2781,6 +2781,8 @@ static inline bool i915_stop_ring_allow_warn(struct drm_i915_private *dev_priv)
> }
>
> void i915_gem_reset(struct drm_device *dev);
> +void i915_gem_reset_engine(struct drm_i915_private *dev_priv,
> + struct intel_engine_cs *engine);
> bool i915_gem_clflush_object(struct drm_i915_gem_object *obj, bool force);
> int __must_check i915_gem_object_finish_gpu(struct drm_i915_gem_object *obj);
> int __must_check i915_gem_init(struct drm_device *dev);
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index 8ce363a..4c88e5c 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -2676,6 +2676,20 @@ void i915_gem_reset(struct drm_device *dev)
> i915_gem_restore_fences(dev);
> }
>
> +void i915_gem_reset_engine(struct drm_i915_private *dev_priv,
> + struct intel_engine_cs *engine)
> +{
> + u32 completed_seqno;
> + struct drm_i915_gem_request *req;
> +
> + completed_seqno = engine->get_seqno(engine, false);
> +
> + /* Find pending batch buffers and mark them as such */
> + list_for_each_entry(req, &engine->request_list, list)
> + if (req && (req->seqno > completed_seqno))
> + i915_set_reset_status(dev_priv, req->ctx, false);
While we're here, can you change this function name; it's one of the
worst I've ever seen. How about update_ctx_hang_stats() which is what it
actually does? (Or ctx_hang_stats_update() if you prefer RPN!).
And if the function i915_gem_reset_ring_status() still exists please
change its name too, it doesn't reset the status of a ring!
.Dave.
More information about the Intel-gfx
mailing list