All of lore.kernel.org
 help / color / mirror / Atom feed
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
To: Oscar Mateo <oscar.mateo@intel.com>, intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH] drm/i915/guc: Keep the ctx_pool_vaddr mapped, for easy access
Date: Wed, 15 Feb 2017 14:37:44 +0200	[thread overview]
Message-ID: <1487162264.8490.18.camel@linux.intel.com> (raw)
In-Reply-To: <1486635819-14980-1-git-send-email-oscar.mateo@intel.com>

On to, 2017-02-09 at 02:23 -0800, Oscar Mateo wrote:
> From: Michal Wajdeczko <michal.wajdeczko@intel.com>
> 
> The GuC descriptor is big in size. If we use a local definition of
> guc_desc we have a chance to overflow stack, so avoid it.
> 
> Also, Chris abhors scatterlists :)
> 
> Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>

<SNIP>

> @@ -116,15 +114,8 @@ static int guc_update_doorbell_id(struct intel_guc *guc,
>  	}
>  
>  	/* Update the GuC's idea of the doorbell ID */
> -	len = sg_pcopy_to_buffer(sg->sgl, sg->nents, &desc, sizeof(desc),
> -			     sizeof(desc) * client->ctx_index);
> -	if (len != sizeof(desc))
> -		return -EFAULT;
> -	desc.db_id = new_id;
> -	len = sg_pcopy_from_buffer(sg->sgl, sg->nents, &desc, sizeof(desc),
> -			     sizeof(desc) * client->ctx_index);
> -	if (len != sizeof(desc))
> -		return -EFAULT;
> +	desc = guc->ctx_pool_vaddr + sizeof(*desc) * client->ctx_index;

As Chris mentioned, could use a __get_doorbell_desc helper for this.

> @@ -895,6 +876,13 @@ int i915_guc_submission_init(struct drm_i915_private *dev_priv)
>  		return PTR_ERR(vma);
>  
>  	guc->ctx_pool_vma = vma;
> +
> +	vaddr = i915_gem_object_pin_map(vma->obj, I915_MAP_WB);
> +	if (IS_ERR(vaddr))

After onion teardown added to this function, propagate errors;

		err = ERR_PTR(vaddr);
> +		goto err;
> +
> +	guc->ctx_pool_vaddr = vaddr;
> +

> @@ -983,8 +971,11 @@ void i915_guc_submission_fini(struct  drm_i915_private *dev_priv)
>  	i915_vma_unpin_and_release(&guc->ads_vma);
>  	i915_vma_unpin_and_release(&guc->log.vma);
>  
> -	if (guc->ctx_pool_vma)
> +	if (guc->ctx_pool_vaddr) {
>  		ida_destroy(&guc->ctx_ids);
> +		i915_gem_object_unpin_map(guc->ctx_pool_vma->obj);
> +	}

This converted to unconditional chain of teardown.

> +
>  	i915_vma_unpin_and_release(&guc->ctx_pool_vma);
>  }
>  
> @@ -1040,5 +1031,3 @@ int intel_guc_resume(struct drm_i915_private *dev_priv)
>  
> >  	return intel_guc_send(guc, data, ARRAY_SIZE(data));
>  }
> -
> -
> diff --git a/drivers/gpu/drm/i915/intel_uc.h b/drivers/gpu/drm/i915/intel_uc.h
> index 8a33a46..29ee373 100644
> --- a/drivers/gpu/drm/i915/intel_uc.h
> +++ b/drivers/gpu/drm/i915/intel_uc.h
> @@ -155,6 +155,7 @@ struct intel_guc {
>  
>  	struct i915_vma *ads_vma;
>  	struct i915_vma *ctx_pool_vma;

s/ctx_pool_vma/ctx_pool/ would bring out the dependency explicitly.

> +	void *ctx_pool_vaddr;
>  	struct ida ctx_ids;
>  
>  	struct i915_guc_client *execbuf_client;

This needs a rebase on top of my cleanup, but is very welcome
improvement.

The types of variables are still wild, but not introduced by this
change.

Regards, Joonas
-- 
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

  parent reply	other threads:[~2017-02-15 12:37 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-02-02 15:27 [PATCH] drm/i915/guc: Dynamically alloc GuC descriptor Oscar Mateo
2017-02-02 23:52 ` ✓ Fi.CI.BAT: success for " Patchwork
2017-02-03  7:33 ` [PATCH] " Chris Wilson
2017-02-07  8:55   ` Oscar Mateo
2017-02-07 17:25     ` Chris Wilson
2017-02-07  9:37       ` Oscar Mateo
2017-02-07 20:49         ` Chris Wilson
2017-02-09 10:23           ` [PATCH] drm/i915/guc: Keep the ctx_pool_vaddr mapped, for easy access Oscar Mateo
2017-02-09 21:01             ` Chris Wilson
2017-02-13 16:05               ` Oscar Mateo
2017-02-15 12:37             ` Joonas Lahtinen [this message]
2017-02-16 14:15               ` [PATCH v2] " Oscar Mateo
2017-02-16 14:18                 ` Oscar Mateo
2017-02-16 22:50                 ` Chris Wilson
2017-02-09 18:52 ` ✓ Fi.CI.BAT: success for drm/i915/guc: Dynamically alloc GuC descriptor (rev2) 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=1487162264.8490.18.camel@linux.intel.com \
    --to=joonas.lahtinen@linux.intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=oscar.mateo@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.