From: Wei Wang <tracywwnj@gmail.com>
To: David Miller <davem@davemloft.net>, netdev@vger.kernel.org
Cc: Eric Dumazet <edumazet@google.com>,
Martin KaFai Lau <kafai@fb.com>, Wei Wang <weiwan@google.com>
Subject: [PATCH net-next 06/21] ipv4: take dst->__refcnt when caching dst in fib
Date: Fri, 16 Jun 2017 10:47:29 -0700 [thread overview]
Message-ID: <20170616174744.139688-7-tracywwnj@gmail.com> (raw)
In-Reply-To: <20170616174744.139688-1-tracywwnj@gmail.com>
From: Wei Wang <weiwan@google.com>
In IPv4 routing code, fib_nh and fib_nh_exception can hold pointers
to struct rtable but they never increment dst->__refcnt.
This leads to the need of the dst garbage collector because when user
is done with this dst and calls dst_release(), it can only decrement
dst->__refcnt and can not free the dst even it sees dst->__refcnt
drops from 1 to 0 (unless DST_NOCACHE flag is set) because the routing
code might still hold reference to it.
And when the routing code tries to delete a route, it has to put the
dst to the gc_list if dst->__refcnt is not yet 0 and have a gc thread
running periodically to check on dst->__refcnt and finally to free dst
when refcnt becomes 0.
This patch increments dst->__refcnt when
fib_nh/fib_nh_exception holds reference to this dst and properly release
the dst when fib_nh/fib_nh_exception has been updated with a new dst.
This patch is a preparation in order to fully get rid of dst gc later.
Signed-off-by: Wei Wang <weiwan@google.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
---
net/ipv4/fib_semantics.c | 5 ++++-
net/ipv4/route.c | 19 ++++++++++++++++---
2 files changed, 20 insertions(+), 4 deletions(-)
diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
index 2157dc08c407..53b3e9c2da4c 100644
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -152,6 +152,7 @@ static void rt_fibinfo_free(struct rtable __rcu **rtp)
* free_fib_info_rcu()
*/
+ dst_release(&rt->dst);
dst_free(&rt->dst);
}
@@ -194,8 +195,10 @@ static void rt_fibinfo_free_cpus(struct rtable __rcu * __percpu *rtp)
struct rtable *rt;
rt = rcu_dereference_protected(*per_cpu_ptr(rtp, cpu), 1);
- if (rt)
+ if (rt) {
+ dst_release(&rt->dst);
dst_free(&rt->dst);
+ }
}
free_percpu(rtp);
}
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 0a843ef2b709..3dee0043117e 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -603,11 +603,13 @@ static void fnhe_flush_routes(struct fib_nh_exception *fnhe)
rt = rcu_dereference(fnhe->fnhe_rth_input);
if (rt) {
RCU_INIT_POINTER(fnhe->fnhe_rth_input, NULL);
+ dst_release(&rt->dst);
rt_free(rt);
}
rt = rcu_dereference(fnhe->fnhe_rth_output);
if (rt) {
RCU_INIT_POINTER(fnhe->fnhe_rth_output, NULL);
+ dst_release(&rt->dst);
rt_free(rt);
}
}
@@ -1332,9 +1334,12 @@ static bool rt_bind_exception(struct rtable *rt, struct fib_nh_exception *fnhe,
rt->rt_gateway = daddr;
if (!(rt->dst.flags & DST_NOCACHE)) {
+ dst_hold(&rt->dst);
rcu_assign_pointer(*porig, rt);
- if (orig)
+ if (orig) {
+ dst_release(&orig->dst);
rt_free(orig);
+ }
ret = true;
}
@@ -1357,12 +1362,20 @@ static bool rt_cache_route(struct fib_nh *nh, struct rtable *rt)
}
orig = *p;
+ /* hold dst before doing cmpxchg() to avoid race condition
+ * on this dst
+ */
+ dst_hold(&rt->dst);
prev = cmpxchg(p, orig, rt);
if (prev == orig) {
- if (orig)
+ if (orig) {
+ dst_release(&orig->dst);
rt_free(orig);
- } else
+ }
+ } else {
+ dst_release(&rt->dst);
ret = false;
+ }
return ret;
}
--
2.13.1.518.g3df882009-goog
next prev parent reply other threads:[~2017-06-16 17:48 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-06-16 17:47 [PATCH net-next 00/21] remove dst garbage collector logic Wei Wang
2017-06-16 17:47 ` [PATCH net-next 01/21] ipv6: remove unnecessary dst_hold() in ip6_fragment() Wei Wang
2017-06-16 17:47 ` [PATCH net-next 02/21] udp: call dst_hold_safe() in udp_sk_rx_set_dst() Wei Wang
2017-06-16 19:02 ` David Miller
2017-06-16 19:06 ` Wei Wang
2017-06-16 17:47 ` [PATCH net-next 03/21] net: use loopback dev when generating blackhole route Wei Wang
2017-06-16 17:47 ` [PATCH net-next 04/21] net: introduce DST_NOGC in dst_release() to destroy dst based on refcnt Wei Wang
2017-06-16 17:47 ` [PATCH net-next 05/21] net: introduce a new function dst_dev_put() Wei Wang
2017-06-16 17:47 ` Wei Wang [this message]
2017-06-16 17:47 ` [PATCH net-next 07/21] ipv4: call dst_dev_put() properly Wei Wang
2017-06-16 17:47 ` [PATCH net-next 08/21] ipv4: call dst_hold_safe() properly Wei Wang
2017-06-16 17:47 ` [PATCH net-next 09/21] ipv4: mark DST_NOGC and remove the operation of dst_free() Wei Wang
2017-06-16 17:47 ` [PATCH net-next 10/21] ipv6: take dst->__refcnt for insertion into fib6 tree Wei Wang
2017-06-16 17:47 ` [PATCH net-next 11/21] ipv6: call dst_dev_put() properly Wei Wang
2017-06-17 14:14 ` kbuild test robot
2017-06-16 17:47 ` [PATCH net-next 12/21] ipv6: call dst_hold_safe() properly Wei Wang
2017-06-16 17:47 ` [PATCH net-next 13/21] ipv6: mark DST_NOGC and remove the operation of dst_free() Wei Wang
2017-06-16 17:47 ` [PATCH net-next 14/21] ipv6: get rid of icmp6 dst garbage collector Wei Wang
2017-06-16 17:47 ` [PATCH net-next 15/21] xfrm: take refcnt of dst when creating struct xfrm_dst bundle Wei Wang
2017-06-16 17:47 ` [PATCH net-next 16/21] decnet: take dst->__refcnt when struct dn_route is created Wei Wang
2017-06-17 16:21 ` kbuild test robot
2017-06-16 17:47 ` [PATCH net-next 17/21] net: remove dst gc related code Wei Wang
2017-06-16 17:47 ` [PATCH net-next 18/21] net: remove DST_NOGC flag Wei Wang
2017-06-16 17:47 ` [PATCH net-next 19/21] net: remove DST_NOCACHE flag Wei Wang
2017-06-16 17:47 ` [PATCH net-next 20/21] net: reorder all the dst flags Wei Wang
2017-06-16 17:47 ` [PATCH net-next 21/21] net: add debug atomic_inc_not_zero() in dst_hold() Wei Wang
2017-06-16 19:07 ` [PATCH net-next 00/21] remove dst garbage collector logic David Miller
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=20170616174744.139688-7-tracywwnj@gmail.com \
--to=tracywwnj@gmail.com \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=kafai@fb.com \
--cc=netdev@vger.kernel.org \
--cc=weiwan@google.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).