diff options
| author | Jakub Kicinski <kuba@kernel.org> | 2024-04-23 02:33:10 +0300 |
|---|---|---|
| committer | Jakub Kicinski <kuba@kernel.org> | 2024-04-23 03:15:39 +0300 |
| commit | af046fd169d43ef0d5f8006954fa4b2fc90974af (patch) | |
| tree | d4fcb4f6735b77eb6d80991252e22b3248ac4e14 /drivers/net/xen-netback/netback.c | |
| parent | 65f1df1140aab935c1db68abdc151dddf6fea85a (diff) | |
| parent | 65bada80dec1f2108a751644773b2120bd789934 (diff) | |
| download | linux-af046fd169d43ef0d5f8006954fa4b2fc90974af.tar.xz | |
Merge branch 'for-uring-ubufops' into HEAD
Pavel Begunkov says:
====================
implement io_uring notification (ubuf_info) stacking (net part)
To have per request buffer notifications each zerocopy io_uring send
request allocates a new ubuf_info. However, as an skb can carry only
one uarg, it may force the stack to create many small skbs hurting
performance in many ways.
The patchset implements notification, i.e. an io_uring's ubuf_info
extension, stacking. It attempts to link ubuf_info's into a list,
allowing to have multiple of them per skb.
liburing/examples/send-zerocopy shows up 6 times performance improvement
for TCP with 4KB bytes per send, and levels it with MSG_ZEROCOPY. Without
the patchset it requires much larger sends to utilise all potential.
bytes | before | after (Kqps)
1200 | 195 | 1023
4000 | 193 | 1386
8000 | 154 | 1058
====================
Link: https://lore.kernel.org/all/cover.1713369317.git.asml.silence@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Diffstat (limited to 'drivers/net/xen-netback/netback.c')
| -rw-r--r-- | drivers/net/xen-netback/netback.c | 11 |
1 files changed, 8 insertions, 3 deletions
diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c index 48254fc07d64..5836995d6774 100644 --- a/drivers/net/xen-netback/netback.c +++ b/drivers/net/xen-netback/netback.c @@ -1157,7 +1157,7 @@ static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *s uarg = skb_shinfo(skb)->destructor_arg; /* increase inflight counter to offset decrement in callback */ atomic_inc(&queue->inflight_packets); - uarg->callback(NULL, uarg, true); + uarg->ops->complete(NULL, uarg, true); skb_shinfo(skb)->destructor_arg = NULL; /* Fill the skb with the new (local) frags. */ @@ -1279,8 +1279,9 @@ static int xenvif_tx_submit(struct xenvif_queue *queue) return work_done; } -void xenvif_zerocopy_callback(struct sk_buff *skb, struct ubuf_info *ubuf_base, - bool zerocopy_success) +static void xenvif_zerocopy_callback(struct sk_buff *skb, + struct ubuf_info *ubuf_base, + bool zerocopy_success) { unsigned long flags; pending_ring_idx_t index; @@ -1313,6 +1314,10 @@ void xenvif_zerocopy_callback(struct sk_buff *skb, struct ubuf_info *ubuf_base, xenvif_skb_zerocopy_complete(queue); } +const struct ubuf_info_ops xenvif_ubuf_ops = { + .complete = xenvif_zerocopy_callback, +}; + static inline void xenvif_tx_dealloc_action(struct xenvif_queue *queue) { struct gnttab_unmap_grant_ref *gop; |