On 2020/4/8 下午10:04, Jean-Philippe Brucker wrote:
The mm_exit() op will be removed from the SVA API. When a process dies
and its mm goes away, the IOMMU driver won't notify device drivers
anymore. Drivers should expect to handle a lot more aborted DMA. On the
upside, it does greatly simplify the queue management.
The uacce_mm struct, that tracks all queues bound to an mm, was only
used by the mm_exit() callback. Remove it.
Signed-off-by: Jean-Philippe Brucker <jean-philippe@xxxxxxxxxx>
Thanks Jean for doing this.
Tested-by: Zhangfei Gao <zhangfei.gao@xxxxxxxxxx>
Except one line.
-static void uacce_mm_put(struct uacce_queue *q)
+static void uacce_unbind_queue(struct uacce_queue *q)
{
- struct uacce_mm *uacce_mm = q->uacce_mm;
-
- lockdep_assert_held(&q->uacce->mm_lock);
-
- mutex_lock(&uacce_mm->lock);
- list_del(&q->list);
- mutex_unlock(&uacce_mm->lock);
-
- if (list_empty(&uacce_mm->queues)) {
- if (uacce_mm->handle)
- iommu_sva_unbind_device(uacce_mm->handle);
- list_del(&uacce_mm->list);
- kfree(uacce_mm);
- }
+ if (!q->handle)
+ return;
+ iommu_sva_unbind_device(q->handle);
+ q->handle = 0;
Otherwise iommu_sva_unbind_device maybe called twice.
Since uacce_unbind_queue can be called by uacce_remove and uacce_fops_release.
Thanks
