On 12/1/21 19:20, Jonathan Lemon wrote:
On Tue, Nov 30, 2021 at 03:18:53PM +0000, Pavel Begunkov wrote:
get_page() in __zerocopy_sg_from_bvec() and matching put_page()s are
expensive. However, we can avoid it if the caller can guarantee that
pages stay alive until the corresponding ubuf_info is not released.
In particular, it targets io_uring with fixed buffers following the
described contract.
Assuming that nobody yet uses bvec together with zerocopy, make all
calls with bvec iterators follow this model.
Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx>
---
[...]
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index b23db60ea6f9..b7b087815539 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -666,10 +666,14 @@ static void skb_release_data(struct sk_buff *skb)
&shinfo->dataref))
goto exit;
- skb_zcopy_clear(skb, true);
+ if (!(shinfo->flags & SKBFL_MANAGED_FRAGS)) {
+ for (i = 0; i < shinfo->nr_frags; i++)
+ __skb_frag_unref(&shinfo->frags[i], skb->pp_recycle);
+ } else {
+ shinfo->flags &= ~SKBFL_MANAGED_FRAGS;
+ }
- for (i = 0; i < shinfo->nr_frags; i++)
- __skb_frag_unref(&shinfo->frags[i], skb->pp_recycle);
+ skb_zcopy_clear(skb, true);
It would be better if all of this logic was moved into the callback
under zcopy_clear. Note that this path is called for both TX and RX.
Yeah, can be done and removes the extra if from non-zc path.
Thanks!
--
Pavel Begunkov
