AMD-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Matthew Auld <matthew.auld@intel.com>
To: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>,
	dri-devel@lists.freedesktop.org, amd-gfx@lists.freedesktop.org,
	intel-gfx@lists.freedesktop.org, nouveau@lists.freedesktop.org
Cc: alexander.deucher@amd.com, luben.tuikov@amd.com,
	christian.koenig@amd.com
Subject: Re: [PATCH v2 4/6] drm/i915: Implement intersect/compatible functions
Date: Tue, 26 Jul 2022 11:41:27 +0100	[thread overview]
Message-ID: <41eea55c-93d6-6406-d279-0583e17ab56c@intel.com> (raw)
In-Reply-To: <20220725114240.4844-4-Arunpravin.PaneerSelvam@amd.com>

On 25/07/2022 12:42, Arunpravin Paneer Selvam wrote:
> Implemented a new intersect and compatible callback function
> fetching start offset from drm buddy allocator.
> 
> Signed-off-by: Christian König <christian.koenig@amd.com>
> Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
> ---
>   drivers/gpu/drm/i915/i915_ttm_buddy_manager.c | 43 +++++++++++++++++++
>   1 file changed, 43 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/i915_ttm_buddy_manager.c b/drivers/gpu/drm/i915/i915_ttm_buddy_manager.c
> index a5109548abc0..b5801c05bd41 100644
> --- a/drivers/gpu/drm/i915/i915_ttm_buddy_manager.c
> +++ b/drivers/gpu/drm/i915/i915_ttm_buddy_manager.c
> @@ -178,6 +178,47 @@ static void i915_ttm_buddy_man_free(struct ttm_resource_manager *man,
>   	kfree(bman_res);
>   }
>   
> +static bool i915_ttm_buddy_man_intersect(struct ttm_resource_manager *man,
> +					 struct ttm_resource *res,
> +					 const struct ttm_place *place,
> +					 size_t size)
> +{
> +	struct i915_ttm_buddy_resource *bman_res = to_ttm_buddy_resource(res);
> +	u32 start, num_pages = PFN_UP(size);
> +	struct drm_buddy_block *block;
> +
> +	/* Check each drm buddy block individually */
> +	list_for_each_entry(block, &bman_res->blocks, link) {
> +		start = drm_buddy_block_offset(block) >> PAGE_SHIFT;
> +		/* Don't evict BOs outside of the requested placement range */
> +		if (place->fpfn >= (start + num_pages) ||
> +		    (place->lpfn && place->lpfn <= start))
> +			return false;
> +	}
> +
> +	return true;
> +}

This looks like a nice idea. We should be able to clean up 
i915_ttm_eviction_valuable() a fair bit I think, if we now call 
ttm_bo_eviction_valuable() at the end (like in amdgpu), and move the 
bits that are specific to buddy_man here?

So something like:

if (!place->fpfn && !place->lpfn)
     return true;

if (!place->fpfn &&
     place->lpfn == i915_buddy_man_visible_size(man))
     return bman_res->used_visible_size > 0;

/* Check each drm buddy block individually */
....

> +
> +static bool i915_ttm_buddy_man_compatible(struct ttm_resource_manager *man,
> +					  struct ttm_resource *res,
> +					  const struct ttm_place *place,
> +					  size_t size)

Is it not possible to derive the size from res->num_pages?

> +{
> +	struct i915_ttm_buddy_resource *bman_res = to_ttm_buddy_resource(res);
> +	u32 start, num_pages = PFN_UP(size);
> +	struct drm_buddy_block *block;
> +
> +	/* Check each drm buddy block individually */
> +	list_for_each_entry(block, &bman_res->blocks, link) {
> +		start = drm_buddy_block_offset(block) >> PAGE_SHIFT;
> +		if (start < place->fpfn ||
> +		    (place->lpfn && (start + num_pages) > place->lpfn))
> +			return false;
> +	}

if (!place->fpfn && !place->lpfn)
     return true;

if (!place->fpfn &&
     place->lpfn == i915_buddy_man_visible_size(man))
     return bman_res->used_visible_size == res->num_pages;

/* Check each drm buddy block individually */
...

> +
> +	return true;
> +}
> +
>   static void i915_ttm_buddy_man_debug(struct ttm_resource_manager *man,
>   				     struct drm_printer *printer)
>   {
> @@ -205,6 +246,8 @@ static void i915_ttm_buddy_man_debug(struct ttm_resource_manager *man,
>   static const struct ttm_resource_manager_func i915_ttm_buddy_manager_func = {
>   	.alloc = i915_ttm_buddy_man_alloc,
>   	.free = i915_ttm_buddy_man_free,
> +	.intersect = i915_ttm_buddy_man_intersect,

s/intersect/intersects/ ?

> +	.compatible = i915_ttm_buddy_man_compatible,
>   	.debug = i915_ttm_buddy_man_debug,
>   };
>   

  reply	other threads:[~2022-07-26 10:41 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-07-25 11:42 [PATCH v2 1/6] drm/ttm: Add new callbacks to ttm res mgr Arunpravin Paneer Selvam
2022-07-25 11:42 ` [PATCH v2 2/6] drm/ttm: Implement intersect/compatible functions Arunpravin Paneer Selvam
2022-08-08 11:30   ` Christian König
2022-08-09  8:26     ` Arunpravin Paneer Selvam
2022-07-25 11:42 ` [PATCH v2 3/6] drm/amdgpu: " Arunpravin Paneer Selvam
2022-07-25 11:42 ` [PATCH v2 4/6] drm/i915: " Arunpravin Paneer Selvam
2022-07-26 10:41   ` Matthew Auld [this message]
2022-07-28 14:36     ` Arunpravin Paneer Selvam
2022-07-25 11:42 ` [PATCH v2 5/6] drm/nouveau: " Arunpravin Paneer Selvam
2022-07-25 11:42 ` [PATCH v2 6/6] drm/ttm: Switch to using the new res callback Arunpravin Paneer Selvam
2022-08-08 11:28 ` [PATCH v2 1/6] drm/ttm: Add new callbacks to ttm res mgr Christian König

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=41eea55c-93d6-6406-d279-0583e17ab56c@intel.com \
    --to=matthew.auld@intel.com \
    --cc=Arunpravin.PaneerSelvam@amd.com \
    --cc=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=christian.koenig@amd.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=luben.tuikov@amd.com \
    --cc=nouveau@lists.freedesktop.org \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox