Re: [Linaro-mm-sig] [PATCH] dma-buf/poll: Get a file reference for outstanding fence callbacks

From: Christian König
Date: Thu Nov 04 2021 - 04:20:57 EST

Am 03.11.21 um 15:50 schrieb Michel Dänzer:
On 2021-07-23 10:22, Christian König wrote:
Am 23.07.21 um 10:19 schrieb Michel Dänzer:
On 2021-07-23 10:04 a.m., Christian König wrote:
Am 23.07.21 um 09:58 schrieb Michel Dänzer:
From: Michel Dänzer <mdaenzer@xxxxxxxxxx>

This makes sure we don't hit the

     BUG_ON(dmabuf->cb_in.active || dmabuf->cb_out.active);

in dma_buf_release, which could be triggered by user space closing the
dma-buf file description while there are outstanding fence callbacks
from dma_buf_poll.
I was also wondering the same thing while working on this, but then thought that the poll interface would take care of this.
I was able to hit the BUG_ON with https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgitlab.gnome.org%2FGNOME%2Fmutter%2F-%2Fmerge_requests%2F1880&amp;data=04%7C01%7Cchristian.koenig%40amd.com%7C8d930ab39011481a839c08d99ed95755%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637715479787056688%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&amp;sdata=SjxSZIsWkP7ru1iHyL0IY9hN9882ENv7Cy38vzOtqyc%3D&amp;reserved=0 .


Cc: stable@xxxxxxxxxxxxxxx
Signed-off-by: Michel Dänzer <mdaenzer@xxxxxxxxxx>
---
   drivers/dma-buf/dma-buf.c | 18 ++++++++++++------
   1 file changed, 12 insertions(+), 6 deletions(-)

diff --git a/drivers/dma-buf/dma-buf.c b/drivers/dma-buf/dma-buf.c
index 6c520c9bd93c..ec25498a971f 100644
--- a/drivers/dma-buf/dma-buf.c
+++ b/drivers/dma-buf/dma-buf.c
@@ -65,12 +65,9 @@ static void dma_buf_release(struct dentry *dentry)
       BUG_ON(dmabuf->vmapping_counter);
         /*
-     * Any fences that a dma-buf poll can wait on should be signaled
-     * before releasing dma-buf. This is the responsibility of each
-     * driver that uses the reservation objects.
-     *
-     * If you hit this BUG() it means someone dropped their ref to the
-     * dma-buf while still having pending operation to the buffer.
+     * If you hit this BUG() it could mean:
+     * * There's a file reference imbalance in dma_buf_poll / dma_buf_poll_cb or somewhere else
+     * * dmabuf->cb_in/out.active are non-0 despite no pending fence callback
        */
       BUG_ON(dmabuf->cb_in.active || dmabuf->cb_out.active);
   @@ -196,6 +193,7 @@ static loff_t dma_buf_llseek(struct file *file, loff_t offset, int whence)
   static void dma_buf_poll_cb(struct dma_fence *fence, struct dma_fence_cb *cb)
   {
       struct dma_buf_poll_cb_t *dcb = (struct dma_buf_poll_cb_t *)cb;
+    struct dma_buf *dmabuf = container_of(dcb->poll, struct dma_buf, poll);
       unsigned long flags;
         spin_lock_irqsave(&dcb->poll->lock, flags);
@@ -203,6 +201,8 @@ static void dma_buf_poll_cb(struct dma_fence *fence, struct dma_fence_cb *cb)
       dcb->active = 0;
       spin_unlock_irqrestore(&dcb->poll->lock, flags);
       dma_fence_put(fence);
+    /* Paired with get_file in dma_buf_poll */
+    fput(dmabuf->file);
Is calling fput() in interrupt context ok? IIRC that could potentially sleep.
Looks fine AFAICT: It has

        if (likely(!in_interrupt() && !(task->flags & PF_KTHREAD))) {

and as a fallback for that, it adds the file to a lock-less delayed_fput_list which is processed by a workqueue.
Ah, yes that makes sense.

Fell free to add Reviewed-by: Christian König <christian.koenig@xxxxxxx>
Thanks! AFAICT this fix can be merged now for 5.16?

I've just pushed it to drm-misc-next-fixes since it won't even apply to drm-misc-fixes.

Could be that we get requests to backport this because of the CC stable.

Christian.