From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
stable@vger.kernel.org,
Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>,
Jason Wang <jasowang@redhat.com>,
"Michael S. Tsirkin" <mst@redhat.com>,
"David S. Miller" <davem@davemloft.net>
Subject: [PATCH 4.20 33/80] virtio_net: Differentiate sk_buff and xdp_frame on freeing
Date: Mon, 4 Feb 2019 11:36:53 +0100 [thread overview]
Message-ID: <20190204103624.585349909@linuxfoundation.org> (raw)
In-Reply-To: <20190204103620.287366543@linuxfoundation.org>
4.20-stable review patch. If anyone has any objections, please let me know.
------------------
From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
[ Upstream commit 5050471d35d1316ba32dfcbb409978337eb9e75e
I had to fold commit df133f3f9625 ("virtio_net: bulk free tx skbs")
into this to make it work. ]
We do not reset or free up unused buffers when enabling/disabling XDP,
so it can happen that xdp_frames are freed after disabling XDP or
sk_buffs are freed after enabling XDP on xdp tx queues.
Thus we need to handle both forms (xdp_frames and sk_buffs) regardless
of XDP setting.
One way to trigger this problem is to disable XDP when napi_tx is
enabled. In that case, virtnet_xdp_set() calls virtnet_napi_enable()
which kicks NAPI. The NAPI handler will call virtnet_poll_cleantx()
which invokes free_old_xmit_skbs() for queues which have been used by
XDP.
Note that even with this change we need to keep skipping
free_old_xmit_skbs() from NAPI handlers when XDP is enabled, because XDP
tx queues do not aquire queue locks.
- v2: Use napi_consume_skb() instead of dev_consume_skb_any()
Fixes: 4941d472bf95 ("virtio-net: do not reset during XDP set")
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/net/virtio_net.c | 64 ++++++++++++++++++++++++++++++++++-------------
1 file changed, 47 insertions(+), 17 deletions(-)
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -57,6 +57,8 @@ module_param(napi_tx, bool, 0644);
#define VIRTIO_XDP_TX BIT(0)
#define VIRTIO_XDP_REDIR BIT(1)
+#define VIRTIO_XDP_FLAG BIT(0)
+
/* RX packet size EWMA. The average packet size is used to determine the packet
* buffer size when refilling RX rings. As the entire RX ring may be refilled
* at once, the weight is chosen so that the EWMA will be insensitive to short-
@@ -251,6 +253,21 @@ struct padded_vnet_hdr {
char padding[4];
};
+static bool is_xdp_frame(void *ptr)
+{
+ return (unsigned long)ptr & VIRTIO_XDP_FLAG;
+}
+
+static void *xdp_to_ptr(struct xdp_frame *ptr)
+{
+ return (void *)((unsigned long)ptr | VIRTIO_XDP_FLAG);
+}
+
+static struct xdp_frame *ptr_to_xdp(void *ptr)
+{
+ return (struct xdp_frame *)((unsigned long)ptr & ~VIRTIO_XDP_FLAG);
+}
+
/* Converting between virtqueue no. and kernel tx/rx queue no.
* 0:rx0 1:tx0 2:rx1 3:tx1 ... 2N:rxN 2N+1:txN 2N+2:cvq
*/
@@ -461,7 +478,8 @@ static int __virtnet_xdp_xmit_one(struct
sg_init_one(sq->sg, xdpf->data, xdpf->len);
- err = virtqueue_add_outbuf(sq->vq, sq->sg, 1, xdpf, GFP_ATOMIC);
+ err = virtqueue_add_outbuf(sq->vq, sq->sg, 1, xdp_to_ptr(xdpf),
+ GFP_ATOMIC);
if (unlikely(err))
return -ENOSPC; /* Caller handle free/refcnt */
@@ -481,13 +499,13 @@ static int virtnet_xdp_xmit(struct net_d
{
struct virtnet_info *vi = netdev_priv(dev);
struct receive_queue *rq = vi->rq;
- struct xdp_frame *xdpf_sent;
struct bpf_prog *xdp_prog;
struct send_queue *sq;
unsigned int len;
int drops = 0;
int kicks = 0;
int ret, err;
+ void *ptr;
int i;
/* Only allow ndo_xdp_xmit if XDP is loaded on dev, as this
@@ -506,8 +524,12 @@ static int virtnet_xdp_xmit(struct net_d
}
/* Free up any pending old buffers before queueing new ones. */
- while ((xdpf_sent = virtqueue_get_buf(sq->vq, &len)) != NULL)
- xdp_return_frame(xdpf_sent);
+ while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) {
+ if (likely(is_xdp_frame(ptr)))
+ xdp_return_frame(ptr_to_xdp(ptr));
+ else
+ napi_consume_skb(ptr, false);
+ }
for (i = 0; i < n; i++) {
struct xdp_frame *xdpf = frames[i];
@@ -1326,20 +1348,28 @@ static int virtnet_receive(struct receiv
return stats.packets;
}
-static void free_old_xmit_skbs(struct send_queue *sq)
+static void free_old_xmit_skbs(struct send_queue *sq, bool in_napi)
{
- struct sk_buff *skb;
unsigned int len;
unsigned int packets = 0;
unsigned int bytes = 0;
+ void *ptr;
- while ((skb = virtqueue_get_buf(sq->vq, &len)) != NULL) {
- pr_debug("Sent skb %p\n", skb);
+ while ((ptr = virtqueue_get_buf(sq->vq, &len)) != NULL) {
+ if (likely(!is_xdp_frame(ptr))) {
+ struct sk_buff *skb = ptr;
- bytes += skb->len;
- packets++;
+ pr_debug("Sent skb %p\n", skb);
- dev_consume_skb_any(skb);
+ bytes += skb->len;
+ napi_consume_skb(skb, in_napi);
+ } else {
+ struct xdp_frame *frame = ptr_to_xdp(ptr);
+
+ bytes += frame->len;
+ xdp_return_frame(frame);
+ }
+ packets++;
}
/* Avoid overhead when no packets have been processed
@@ -1375,7 +1405,7 @@ static void virtnet_poll_cleantx(struct
return;
if (__netif_tx_trylock(txq)) {
- free_old_xmit_skbs(sq);
+ free_old_xmit_skbs(sq, true);
__netif_tx_unlock(txq);
}
@@ -1459,7 +1489,7 @@ static int virtnet_poll_tx(struct napi_s
txq = netdev_get_tx_queue(vi->dev, index);
__netif_tx_lock(txq, raw_smp_processor_id());
- free_old_xmit_skbs(sq);
+ free_old_xmit_skbs(sq, true);
__netif_tx_unlock(txq);
virtqueue_napi_complete(napi, sq->vq, 0);
@@ -1528,7 +1558,7 @@ static netdev_tx_t start_xmit(struct sk_
bool use_napi = sq->napi.weight;
/* Free up any pending old buffers before queueing new ones. */
- free_old_xmit_skbs(sq);
+ free_old_xmit_skbs(sq, false);
if (use_napi && kick)
virtqueue_enable_cb_delayed(sq->vq);
@@ -1571,7 +1601,7 @@ static netdev_tx_t start_xmit(struct sk_
if (!use_napi &&
unlikely(!virtqueue_enable_cb_delayed(sq->vq))) {
/* More just got used, free them then recheck. */
- free_old_xmit_skbs(sq);
+ free_old_xmit_skbs(sq, false);
if (sq->vq->num_free >= 2+MAX_SKB_FRAGS) {
netif_start_subqueue(dev, qnum);
virtqueue_disable_cb(sq->vq);
@@ -2639,10 +2669,10 @@ static void free_unused_bufs(struct virt
for (i = 0; i < vi->max_queue_pairs; i++) {
struct virtqueue *vq = vi->sq[i].vq;
while ((buf = virtqueue_detach_unused_buf(vq)) != NULL) {
- if (!is_xdp_raw_buffer_queue(vi, i))
+ if (!is_xdp_frame(buf))
dev_kfree_skb(buf);
else
- xdp_return_frame(buf);
+ xdp_return_frame(ptr_to_xdp(buf));
}
}
next prev parent reply other threads:[~2019-02-04 10:52 UTC|newest]
Thread overview: 84+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-02-04 10:36 [PATCH 4.20 00/80] 4.20.7-stable review Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 01/80] drm/msm/gpu: fix building without debugfs Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 02/80] ipv6: sr: clear IP6CB(skb) on SRH ip4ip6 encapsulation Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 03/80] ipvlan, l3mdev: fix broken l3s mode wrt local routes Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 04/80] l2tp: copy 4 more bytes to linear part if necessary Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 05/80] l2tp: fix reading optional fields of L2TPv3 Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 06/80] net: ip_gre: always reports o_key to userspace Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 07/80] net: ip_gre: use erspan key field for tunnel lookup Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 08/80] net/ipv6: dont return positive numbers when nothing was dumped Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 09/80] net/mlx4_core: Add masking for a few queries on HCA caps Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 10/80] netrom: switch to sock timer API Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 11/80] net/rose: fix NULL ax25_cb kernel panic Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 12/80] net: set default network namespace in init_dummy_netdev() Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 13/80] ravb: expand rx descriptor data to accommodate hw checksum Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 14/80] sctp: improve the events for sctp stream reset Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 15/80] tun: move the call to tun_set_real_num_queues Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 16/80] ucc_geth: Reset BQL queue when stopping device Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 17/80] vhost: fix OOB in get_rx_bufs() Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 18/80] net: ip6_gre: always reports o_key to userspace Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 19/80] sctp: improve the events for sctp stream adding Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 20/80] net/mlx5e: Allow MAC invalidation while spoofchk is ON Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 21/80] ip6mr: Fix notifiers call on mroute_clean_tables() Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 22/80] Revert "net/mlx5e: E-Switch, Initialize eswitch only if eswitch manager" Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 23/80] sctp: set chunk transport correctly when its a new asoc Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 24/80] sctp: set flow sport from saddr only when its 0 Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 25/80] net: tls: Fix deadlock in free_resources tx Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 26/80] net: tls: Save iv in tls_rec for async crypto requests Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 27/80] virtio_net: Dont enable NAPI when interface is down Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 28/80] virtio_net: Dont call free_old_xmit_skbs for xdp_frames Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 29/80] virtio_net: Fix not restoring real_num_rx_queues Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 30/80] virtio_net: Fix out of bounds access of sq Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 31/80] virtio_net: Dont process redirected XDP frames when XDP is disabled Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 32/80] virtio_net: Use xdp_return_frame to free xdp_frames on destroying vqs Greg Kroah-Hartman
2019-02-04 10:36 ` Greg Kroah-Hartman [this message]
2019-02-04 10:36 ` [PATCH 4.20 34/80] ipv6: Consider sk_bound_dev_if when binding a socket to an address Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 35/80] CIFS: Do not count -ENODATA as failure for query directory Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 36/80] CIFS: Fix possible oops and memory leaks in async IO Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 37/80] CIFS: Fix trace command logging for SMB2 reads and writes Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 38/80] CIFS: fix use-after-free of the lease keys Greg Kroah-Hartman
2019-02-04 10:36 ` [PATCH 4.20 39/80] CIFS: Do not consider -ENODATA as stat failure for reads Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 40/80] fs/dcache: Fix incorrect nr_dentry_unused accounting in shrink_dcache_sb() Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 41/80] iommu/vt-d: Fix memory leak in intel_iommu_put_resv_regions() Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 42/80] selftests/seccomp: Enhance per-arch ptrace syscall skip tests Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 43/80] NFS: Fix up return value on fatal errors in nfs_page_async_flush() Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 44/80] ARM: cns3xxx: Fix writing to wrong PCI config registers after alignment Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 45/80] arm64: kaslr: ensure randomized quantities are clean also when kaslr is off Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 46/80] arm64: Do not issue IPIs for user executable ptes Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 47/80] arm64: hyp-stub: Forbid kprobing of the hyp-stub Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 48/80] arm64: hibernate: Clean the __hyp_text to PoC after resume Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 49/80] gpio: altera-a10sr: Set proper output level for direction_output Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 50/80] gpiolib: fix line event timestamps for nested irqs Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 51/80] gpio: pcf857x: Fix interrupts on multiple instances Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 52/80] gpio: sprd: Fix the incorrect data register Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 53/80] gpio: sprd: Fix incorrect irq type setting for the async EIC Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 54/80] gfs2: Revert "Fix loop in gfs2_rbm_find" Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 55/80] mmc: bcm2835: Fix DMA channel leak on probe error Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 56/80] mmc: mediatek: fix incorrect register setting of hs400_cmd_int_delay Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 57/80] ALSA: usb-audio: Add Opus #3 to quirks for native DSD support Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 58/80] ALSA: hda/realtek - Fixed hp_pin no value Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 59/80] ALSA: pcm: Fix tight loop of OSS capture stream Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 60/80] IB/uverbs: Fix OOPs upon device disassociation Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 61/80] IB/uverbs: Fix OOPs in uverbs_user_mmap_disassociate Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 62/80] IB/hfi1: Remove overly conservative VM_EXEC flag check Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 63/80] IB/hfi1: Add limit test for RC/UC send via loopback Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 64/80] platform/x86: asus-nb-wmi: Map 0x35 to KEY_SCREENLOCK Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 65/80] platform/x86: asus-nb-wmi: Drop mapping of 0x33 and 0x34 scan codes Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 66/80] Btrfs: fix deadlock when allocating tree block during leaf/node split Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 67/80] btrfs: On error always free subvol_name in btrfs_mount Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 68/80] kernel/exit.c: release ptraced tasks before zap_pid_ns_processes Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 69/80] mm/hugetlb.c: teach follow_hugetlb_page() to handle FOLL_NOWAIT Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 70/80] oom, oom_reaper: do not enqueue same task twice Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 71/80] mm,memory_hotplug: fix scan_movable_pages() for gigantic hugepages Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 72/80] mm, oom: fix use-after-free in oom_kill_process Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 73/80] mm: hwpoison: use do_send_sig_info() instead of force_sig() Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 74/80] mm: migrate: dont rely on __PageMovable() of newpage after unlocking it Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 75/80] of: overlay: add tests to validate kfrees from overlay removal Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 76/80] of: overlay: add missing of_node_get() in __of_attach_node_sysfs Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 77/80] of: overlay: use prop add changeset entry for property in new nodes Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 78/80] of: overlay: do not duplicate properties from overlay for " Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 79/80] md/raid5: fix out of memory during raid cache recovery Greg Kroah-Hartman
2019-02-04 10:37 ` [PATCH 4.20 80/80] cifs: Always resolve hostname before reconnecting Greg Kroah-Hartman
2019-02-04 21:49 ` [PATCH 4.20 00/80] 4.20.7-stable review Guenter Roeck
2019-02-05 6:20 ` Naresh Kamboju
2019-02-05 14:43 ` Greg Kroah-Hartman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190204103624.585349909@linuxfoundation.org \
--to=gregkh@linuxfoundation.org \
--cc=davem@davemloft.net \
--cc=jasowang@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=makita.toshiaki@lab.ntt.co.jp \
--cc=mst@redhat.com \
--cc=stable@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).