dri-devel.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: David Herrmann <dh.herrmann@gmail.com>
To: dri-devel@lists.freedesktop.org
Cc: Dave Airlie <airlied@redhat.com>, Daniel Vetter <daniel.vetter@ffwll.ch>
Subject: [PATCH 2/4] drm/ttm: replace drm_mm_pre_get() by direct alloc
Date: Thu, 25 Jul 2013 15:56:00 +0200	[thread overview]
Message-ID: <1374760562-6096-3-git-send-email-dh.herrmann@gmail.com> (raw)
In-Reply-To: <1374760562-6096-1-git-send-email-dh.herrmann@gmail.com>

Instead of calling drm_mm_pre_get() in a row, we now preallocate the node
and then use the atomic insertion functions. This has the exact same
semantics and there is no reason to use the racy pre-allocations.

Note that ttm_bo_man_get_node() does not run in atomic context. Nouveau already
uses GFP_KERNEL alloc in nouveau/nouveau_ttm.c in nouveau_gart_manager_new(). So
we can do the same in ttm_bo_man_get_node().

Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
---
 drivers/gpu/drm/ttm/ttm_bo_manager.c | 40 +++++++++++++++++-------------------
 1 file changed, 19 insertions(+), 21 deletions(-)

diff --git a/drivers/gpu/drm/ttm/ttm_bo_manager.c b/drivers/gpu/drm/ttm/ttm_bo_manager.c
index e4367f9..cbd2ec7 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_manager.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_manager.c
@@ -61,28 +61,24 @@ static int ttm_bo_man_get_node(struct ttm_mem_type_manager *man,
 	lpfn = placement->lpfn;
 	if (!lpfn)
 		lpfn = man->size;
-	do {
-		ret = drm_mm_pre_get(mm);
-		if (unlikely(ret))
-			return ret;
 
-		spin_lock(&rman->lock);
-		node = drm_mm_search_free_in_range(mm,
-					mem->num_pages, mem->page_alignment,
-					placement->fpfn, lpfn, 1);
-		if (unlikely(node == NULL)) {
-			spin_unlock(&rman->lock);
-			return 0;
-		}
-		node = drm_mm_get_block_atomic_range(node, mem->num_pages,
-						     mem->page_alignment,
-						     placement->fpfn,
-						     lpfn);
-		spin_unlock(&rman->lock);
-	} while (node == NULL);
+	node = kzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node)
+		return -ENOMEM;
+
+	spin_lock(&rman->lock);
+	ret = drm_mm_insert_node_in_range(mm, node, mem->num_pages,
+					  mem->page_alignment,
+					  placement->fpfn, lpfn, true);
+	spin_unlock(&rman->lock);
+
+	if (unlikely(ret)) {
+		kfree(node);
+	} else {
+		mem->mm_node = node;
+		mem->start = node->start;
+	}
 
-	mem->mm_node = node;
-	mem->start = node->start;
 	return 0;
 }
 
@@ -93,8 +89,10 @@ static void ttm_bo_man_put_node(struct ttm_mem_type_manager *man,
 
 	if (mem->mm_node) {
 		spin_lock(&rman->lock);
-		drm_mm_put_block(mem->mm_node);
+		drm_mm_remove_node(mem->mm_node);
 		spin_unlock(&rman->lock);
+
+		kfree(mem->mm_node);
 		mem->mm_node = NULL;
 	}
 }
-- 
1.8.3.3

  parent reply	other threads:[~2013-07-25 13:56 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-07-25 13:55 [PATCH 0/4] DRM: Remove MM "Pre-Alloc" David Herrmann
2013-07-25 13:55 ` [PATCH 1/4] drm/mm: add "best_match" to drm_mm_insert_node() David Herrmann
2013-07-25 14:15   ` Daniel Vetter
2013-07-27 11:36   ` [PATCH 1/4] drm/mm: add "best_match" flag " David Herrmann
2013-08-05  7:46     ` Daniel Vetter
2013-08-06  9:39       ` David Herrmann
2013-08-06 11:03         ` Daniel Vetter
2013-07-25 13:56 ` David Herrmann [this message]
2013-07-25 14:25   ` [PATCH 2/4] drm/ttm: replace drm_mm_pre_get() by direct alloc Daniel Vetter
2013-07-27 11:37   ` [PATCH v2 " David Herrmann
2013-07-25 13:56 ` [PATCH 3/4] drm/i915: pre-alloc instead of drm_mm search/get_block David Herrmann
2013-07-25 14:14   ` Chris Wilson
2013-07-27 11:38   ` [PATCH v2 " David Herrmann
2013-07-27 13:06     ` Chris Wilson
2013-07-27 13:09       ` David Herrmann
2013-07-27 13:15         ` Chris Wilson
2013-07-27 14:21     ` [PATCH v3 " David Herrmann
2013-08-05  7:49       ` Daniel Vetter
2013-07-25 13:56 ` [PATCH 4/4] drm/mm: remove unused API David Herrmann
2013-07-25 14:27   ` Daniel Vetter
2013-07-27 11:39   ` [PATCH v2 " David Herrmann

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=1374760562-6096-3-git-send-email-dh.herrmann@gmail.com \
    --to=dh.herrmann@gmail.com \
    --cc=airlied@redhat.com \
    --cc=daniel.vetter@ffwll.ch \
    --cc=dri-devel@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;
as well as URLs for NNTP newsgroup(s).