Hi guys,
we are currently working an Freesync and direct scan out from system
memory on AMD APUs in A+A laptops.
On problem we stumbled over is that our display hardware needs to scan
out from uncached system memory and we currently don't have a way to
communicate that through DMA-buf.
For our specific use case at hand we are going to implement something
driver specific, but the question is should we have something more
generic for this?
After all the system memory access pattern is a PCIe extension and as
such something generic.
Regards,
Christian.
krealloc_array() ignores attempts to reduce the array size, so the attempt
to save memory is completely pointless here.
Also move testing for the no fence case into sync_file_set_fence(), this
way we don't even touch the fence array when we don't have any fences.
Signed-off-by: Christian König <christian.koenig(a)amd.com>
---
drivers/dma-buf/sync_file.c | 21 +++++++--------------
1 file changed, 7 insertions(+), 14 deletions(-)
diff --git a/drivers/dma-buf/sync_file.c b/drivers/dma-buf/sync_file.c
index 514d213261df..0fe564539166 100644
--- a/drivers/dma-buf/sync_file.c
+++ b/drivers/dma-buf/sync_file.c
@@ -157,9 +157,15 @@ static int sync_file_set_fence(struct sync_file *sync_file,
* we already own a new reference to the fence. For num_fence > 1
* we own the reference of the dma_fence_array creation.
*/
- if (num_fences == 1) {
+
+ if (num_fences == 0) {
+ sync_file->fence = dma_fence_get_stub();
+ kfree(fences);
+
+ } else if (num_fences == 1) {
sync_file->fence = fences[0];
kfree(fences);
+
} else {
array = dma_fence_array_create(num_fences, fences,
dma_fence_context_alloc(1),
@@ -261,19 +267,6 @@ static struct sync_file *sync_file_merge(const char *name, struct sync_file *a,
}
}
- if (index == 0)
- fences[index++] = dma_fence_get_stub();
-
- if (num_fences > index) {
- struct dma_fence **tmp;
-
- /* Keep going even when reducing the size failed */
- tmp = krealloc_array(fences, index, sizeof(*fences),
- GFP_KERNEL);
- if (tmp)
- fences = tmp;
- }
-
if (sync_file_set_fence(sync_file, fences, index) < 0)
goto err_put_fences;
--
2.25.1
'cat /sys/kernel/debug/dma_buf/bufinfo' will print the Dma-buf
Objects' information when the CONFIG_DEBUG_FS=y.
However, the printed table header information does not contain
the name field. So we need to add the name field to the table
header and use the '<none>' to replace the empty buf_obj->name.
Signed-off-by: Yuanzheng Song <songyuanzheng(a)huawei.com>
---
drivers/dma-buf/dma-buf.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/dma-buf.c b/drivers/dma-buf/dma-buf.c
index 79795857be3e..a2f9a1815e38 100644
--- a/drivers/dma-buf/dma-buf.c
+++ b/drivers/dma-buf/dma-buf.c
@@ -1351,7 +1351,7 @@ static int dma_buf_debug_show(struct seq_file *s, void *unused)
return ret;
seq_puts(s, "\nDma-buf Objects:\n");
- seq_printf(s, "%-8s\t%-8s\t%-8s\t%-8s\texp_name\t%-8s\n",
+ seq_printf(s, "%-8s\t%-8s\t%-8s\t%-8s\texp_name\t%-8s\tname\n",
"size", "flags", "mode", "count", "ino");
list_for_each_entry(buf_obj, &db_list.head, list_node) {
@@ -1368,7 +1368,7 @@ static int dma_buf_debug_show(struct seq_file *s, void *unused)
file_count(buf_obj->file),
buf_obj->exp_name,
file_inode(buf_obj->file)->i_ino,
- buf_obj->name ?: "");
+ buf_obj->name ?: "<none>");
spin_unlock(&buf_obj->name_lock);
dma_resv_describe(buf_obj->resv, s);
--
2.25.1
