Re: [PATCH 2/2] drm/ttm: don't write to bo-reserved without holding glob-lru_lock
On Mon, Mar 01, 2010 at 07:34:40PM +0100, Maarten Maathuis wrote:
- The headerfile says you can't write to it without holding the lock.
Signed-off-by: Maarten Maathuis madman2...@gmail.com
NAK, from my POV as we always use atomic_* on reserved it's useless
to protect it with spinlock.
Cheers,
Jerome
---
drivers/gpu/drm/ttm/ttm_bo.c |8 +++-
1 files changed, 7 insertions(+), 1 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index f5333d9..2104885 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -476,9 +476,9 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object
*bo, bool remove_all)
drm_mm_put_block(bo-mem.mm_node);
bo-mem.mm_node = NULL;
}
- spin_unlock(glob-lru_lock);
atomic_set(bo-reserved, 0);
+ spin_unlock(glob-lru_lock);
while (put_count--)
kref_put(bo-list_kref, ttm_bo_ref_bug);
@@ -1707,8 +1707,12 @@ EXPORT_SYMBOL(ttm_bo_wait);
void ttm_bo_unblock_reservation(struct ttm_buffer_object *bo)
{
+ struct ttm_bo_global *glob = bo-glob;
+
+ spin_lock(glob-lru_lock);
atomic_set(bo-reserved, 0);
wake_up_all(bo-event_queue);
+ spin_unlock(glob-lru_lock);
}
int ttm_bo_block_reservation(struct ttm_buffer_object *bo, bool
interruptible,
@@ -1849,8 +1853,10 @@ out:
* already swapped buffer.
*/
+ spin_lock(glob-lru_lock);
atomic_set(bo-reserved, 0);
wake_up_all(bo-event_queue);
+ spin_unlock(glob-lru_lock);
kref_put(bo-list_kref, ttm_bo_release_list);
return ret;
}
--
1.7.0
--
Download Intel#174; Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
--
___
Dri-devel mailing list
Dri-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dri-devel
--
Download Intel#174; Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
--
___
Dri-devel mailing list
Dri-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: [PATCH 2/2] drm/ttm: don't write to bo-reserved without holding glob-lru_lock
Maarten Maathuis wrote:
- The headerfile says you can't write to it without holding the lock.
NAK.
The header-file is incorrect. It should say that the lru_lock needs to
be held only when we transition from 0 to 1.
That guarantees that if the lru lock is held, and we read 0, we can
safely assume that we're the only ones trying to lock.
/Thomas
Signed-off-by: Maarten Maathuis madman2...@gmail.com
---
drivers/gpu/drm/ttm/ttm_bo.c |8 +++-
1 files changed, 7 insertions(+), 1 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index f5333d9..2104885 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -476,9 +476,9 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object
*bo, bool remove_all)
drm_mm_put_block(bo-mem.mm_node);
bo-mem.mm_node = NULL;
}
- spin_unlock(glob-lru_lock);
atomic_set(bo-reserved, 0);
+ spin_unlock(glob-lru_lock);
while (put_count--)
kref_put(bo-list_kref, ttm_bo_ref_bug);
@@ -1707,8 +1707,12 @@ EXPORT_SYMBOL(ttm_bo_wait);
void ttm_bo_unblock_reservation(struct ttm_buffer_object *bo)
{
+ struct ttm_bo_global *glob = bo-glob;
+
+ spin_lock(glob-lru_lock);
atomic_set(bo-reserved, 0);
wake_up_all(bo-event_queue);
+ spin_unlock(glob-lru_lock);
}
int ttm_bo_block_reservation(struct ttm_buffer_object *bo, bool
interruptible,
@@ -1849,8 +1853,10 @@ out:
* already swapped buffer.
*/
+ spin_lock(glob-lru_lock);
atomic_set(bo-reserved, 0);
wake_up_all(bo-event_queue);
+ spin_unlock(glob-lru_lock);
kref_put(bo-list_kref, ttm_bo_release_list);
return ret;
}
--
Download Intel#174; Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
--
___
Dri-devel mailing list
Dri-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dri-devel
