From: "Björn Töpel" <bjorn.topel@gmail.com>
To: netdev@vger.kernel.org, ast@kernel.org, daniel@iogearbox.net
Cc: "Björn Töpel" <bjorn.topel@intel.com>,
bpf@vger.kernel.org, davem@davemloft.net,
jakub.kicinski@netronome.com, hawk@kernel.org,
john.fastabend@gmail.com, magnus.karlsson@intel.com,
jonathan.lemon@gmail.com,
"Toke Høiland-Jørgensen" <toke@redhat.com>
Subject: [PATCH bpf-next v2 1/8] xdp: simplify devmap cleanup
Date: Thu, 19 Dec 2019 07:09:59 +0100 [thread overview]
Message-ID: <20191219061006.21980-2-bjorn.topel@gmail.com> (raw)
In-Reply-To: <20191219061006.21980-1-bjorn.topel@gmail.com>
From: Björn Töpel <bjorn.topel@intel.com>
After the RCU flavor consolidation [1], call_rcu() and
synchronize_rcu() waits for preempt-disable regions (NAPI) in addition
to the read-side critical sections. As a result of this, the cleanup
code in devmap can be simplified
* There is no longer a need to flush in __dev_map_entry_free, since we
know that this has been done when the call_rcu() callback is
triggered.
* When freeing the map, there is no need to explicitly wait for a
flush. It's guaranteed to be done after the synchronize_rcu() call
in dev_map_free(). The rcu_barrier() is still needed, so that the
map is not freed prior the elements.
[1] https://lwn.net/Articles/777036/
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
---
kernel/bpf/devmap.c | 43 +++++--------------------------------------
1 file changed, 5 insertions(+), 38 deletions(-)
diff --git a/kernel/bpf/devmap.c b/kernel/bpf/devmap.c
index 3d3d61b5985b..b7595de6a91a 100644
--- a/kernel/bpf/devmap.c
+++ b/kernel/bpf/devmap.c
@@ -201,7 +201,7 @@ static struct bpf_map *dev_map_alloc(union bpf_attr *attr)
static void dev_map_free(struct bpf_map *map)
{
struct bpf_dtab *dtab = container_of(map, struct bpf_dtab, map);
- int i, cpu;
+ int i;
/* At this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0,
* so the programs (can be more than one that used this map) were
@@ -221,18 +221,6 @@ static void dev_map_free(struct bpf_map *map)
/* Make sure prior __dev_map_entry_free() have completed. */
rcu_barrier();
- /* To ensure all pending flush operations have completed wait for flush
- * list to empty on _all_ cpus.
- * Because the above synchronize_rcu() ensures the map is disconnected
- * from the program we can assume no new items will be added.
- */
- for_each_online_cpu(cpu) {
- struct list_head *flush_list = per_cpu_ptr(dtab->flush_list, cpu);
-
- while (!list_empty(flush_list))
- cond_resched();
- }
-
if (dtab->map.map_type == BPF_MAP_TYPE_DEVMAP_HASH) {
for (i = 0; i < dtab->n_buckets; i++) {
struct bpf_dtab_netdev *dev;
@@ -345,8 +333,7 @@ static int dev_map_hash_get_next_key(struct bpf_map *map, void *key,
return -ENOENT;
}
-static int bq_xmit_all(struct xdp_bulk_queue *bq, u32 flags,
- bool in_napi_ctx)
+static int bq_xmit_all(struct xdp_bulk_queue *bq, u32 flags)
{
struct bpf_dtab_netdev *obj = bq->obj;
struct net_device *dev = obj->dev;
@@ -384,11 +371,7 @@ static int bq_xmit_all(struct xdp_bulk_queue *bq, u32 flags,
for (i = 0; i < bq->count; i++) {
struct xdp_frame *xdpf = bq->q[i];
- /* RX path under NAPI protection, can return frames faster */
- if (likely(in_napi_ctx))
- xdp_return_frame_rx_napi(xdpf);
- else
- xdp_return_frame(xdpf);
+ xdp_return_frame_rx_napi(xdpf);
drops++;
}
goto out;
@@ -409,7 +392,7 @@ void __dev_map_flush(struct bpf_map *map)
rcu_read_lock();
list_for_each_entry_safe(bq, tmp, flush_list, flush_node)
- bq_xmit_all(bq, XDP_XMIT_FLUSH, true);
+ bq_xmit_all(bq, XDP_XMIT_FLUSH);
rcu_read_unlock();
}
@@ -440,7 +423,7 @@ static int bq_enqueue(struct bpf_dtab_netdev *obj, struct xdp_frame *xdpf,
struct xdp_bulk_queue *bq = this_cpu_ptr(obj->bulkq);
if (unlikely(bq->count == DEV_MAP_BULK_SIZE))
- bq_xmit_all(bq, 0, true);
+ bq_xmit_all(bq, 0);
/* Ingress dev_rx will be the same for all xdp_frame's in
* bulk_queue, because bq stored per-CPU and must be flushed
@@ -509,27 +492,11 @@ static void *dev_map_hash_lookup_elem(struct bpf_map *map, void *key)
return dev ? &dev->ifindex : NULL;
}
-static void dev_map_flush_old(struct bpf_dtab_netdev *dev)
-{
- if (dev->dev->netdev_ops->ndo_xdp_xmit) {
- struct xdp_bulk_queue *bq;
- int cpu;
-
- rcu_read_lock();
- for_each_online_cpu(cpu) {
- bq = per_cpu_ptr(dev->bulkq, cpu);
- bq_xmit_all(bq, XDP_XMIT_FLUSH, false);
- }
- rcu_read_unlock();
- }
-}
-
static void __dev_map_entry_free(struct rcu_head *rcu)
{
struct bpf_dtab_netdev *dev;
dev = container_of(rcu, struct bpf_dtab_netdev, rcu);
- dev_map_flush_old(dev);
free_percpu(dev->bulkq);
dev_put(dev->dev);
kfree(dev);
--
2.20.1
next prev parent reply other threads:[~2019-12-19 6:10 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-12-19 6:09 [PATCH bpf-next v2 0/8] Simplify xdp_do_redirect_map()/xdp_do_flush_map() and XDP maps Björn Töpel
2019-12-19 6:09 ` Björn Töpel [this message]
2020-01-07 17:32 ` [PATCH bpf-next v2 1/8] xdp: simplify devmap cleanup John Fastabend
2020-01-08 10:08 ` Björn Töpel
2019-12-19 6:10 ` [PATCH bpf-next v2 2/8] xdp: simplify cpumap cleanup Björn Töpel
2019-12-19 6:10 ` [PATCH bpf-next v2 3/8] xdp: fix graze->grace type-o in cpumap comments Björn Töpel
2020-01-07 17:33 ` John Fastabend
2019-12-19 6:10 ` [PATCH bpf-next v2 4/8] xsk: make xskmap flush_list common for all map instances Björn Töpel
2020-01-07 17:54 ` John Fastabend
2020-01-08 10:13 ` Björn Töpel
2020-01-08 15:52 ` John Fastabend
2020-01-08 16:01 ` Toke Høiland-Jørgensen
2019-12-19 6:10 ` [PATCH bpf-next v2 5/8] xdp: make devmap " Björn Töpel
2020-01-07 17:58 ` John Fastabend
2020-01-08 10:16 ` Björn Töpel
2020-01-08 10:23 ` Toke Høiland-Jørgensen
2020-01-08 10:25 ` Björn Töpel
2019-12-19 6:10 ` [PATCH bpf-next v2 6/8] xdp: make cpumap " Björn Töpel
2020-01-07 17:59 ` John Fastabend
2019-12-19 6:10 ` [PATCH bpf-next v2 7/8] xdp: remove map_to_flush and map swap detection Björn Töpel
2020-01-07 18:15 ` John Fastabend
2020-01-07 21:07 ` Toke Høiland-Jørgensen
2020-01-08 3:45 ` John Fastabend
2020-01-08 10:24 ` Björn Töpel
2019-12-19 6:10 ` [PATCH bpf-next v2 8/8] xdp: simplify __bpf_tx_xdp_map() Björn Töpel
2019-12-19 7:18 ` [PATCH bpf-next v2 0/8] Simplify xdp_do_redirect_map()/xdp_do_flush_map() and XDP maps Jesper Dangaard Brouer
2019-12-20 5:21 ` Alexei Starovoitov
2019-12-20 7:46 ` Jesper Dangaard Brouer
2019-12-20 9:26 ` Jesper Dangaard Brouer
2019-12-20 10:29 ` Toke Høiland-Jørgensen
2020-01-07 11:10 ` Björn Töpel
2020-01-07 11:25 ` Toke Høiland-Jørgensen
2020-01-07 13:05 ` Jesper Dangaard Brouer
2020-01-07 13:27 ` Toke Høiland-Jørgensen
2020-01-07 13:52 ` Jesper Dangaard Brouer
2020-01-07 14:18 ` Toke Høiland-Jørgensen
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=20191219061006.21980-2-bjorn.topel@gmail.com \
--to=bjorn.topel@gmail.com \
--cc=ast@kernel.org \
--cc=bjorn.topel@intel.com \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
--cc=davem@davemloft.net \
--cc=hawk@kernel.org \
--cc=jakub.kicinski@netronome.com \
--cc=john.fastabend@gmail.com \
--cc=jonathan.lemon@gmail.com \
--cc=magnus.karlsson@intel.com \
--cc=netdev@vger.kernel.org \
--cc=toke@redhat.com \
/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).