Fix a TCP loss recovery performance bug raised recently on the netdev
list, in two threads:
(i) July 26, 2017: netdev thread "TCP fast retransmit issues"
(ii) July 26, 2017: netdev thread:
"[PATCH V2 net-next] TLP: Don't reschedule PTO when there's one
outstanding TLP retransmission"
The basic problem is that incoming TCP packets that did not indicate
forward progress could cause the xmit timer (TLP or RTO) to be rearmed
and pushed back in time. In certain corner cases this could result in
the following problems noted in these threads:
- Repeated ACKs coming in with bogus SACKs corrupted by middleboxes
could cause TCP to repeatedly schedule TLPs forever. We kept
sending TLPs after every ~200ms, which elicited bogus SACKs, which
caused more TLPs, ad infinitum; we never fired an RTO to fill in
the holes.
- Incoming data segments could, in some cases, cause us to reschedule
our RTO or TLP timer further out in time, for no good reason. This
could cause repeated inbound data to result in stalls in outbound
data, in the presence of packet loss.
This commit fixes these bugs by changing the TLP and RTO ACK
processing to:
(a) Only reschedule the xmit timer once per ACK.
(b) Only reschedule the xmit timer if tcp_clean_rtx_queue() deems the
ACK indicates sufficient forward progress (a packet was
cumulatively ACKed, or we got a SACK for a packet that was sent
before the most recent retransmit of the write queue head).
This brings us back into closer compliance with the RFCs, since, as
the comment for tcp_rearm_rto() notes, we should only restart the RTO
timer after forward progress on the connection. Previously we were
restarting the xmit timer even in these cases where there was no
forward progress.
As a side benefit, this commit simplifies and speeds up the TCP timer
arming logic. We had been calling inet_csk_reset_xmit_timer() three
times on normal ACKs that cumulatively acknowledged some data:
1) Once near the top of tcp_ack() to switch from TLP timer to RTO:
if (icsk->icsk_pending == ICSK_TIME_LOSS_PROBE)
tcp_rearm_rto(sk);
2) Once in tcp_clean_rtx_queue(), to update the RTO:
if (flag & FLAG_ACKED) {
tcp_rearm_rto(sk);
3) Once in tcp_ack() after tcp_fastretrans_alert() to switch from RTO
to TLP:
if (icsk->icsk_pending == ICSK_TIME_RETRANS)
tcp_schedule_loss_probe(sk);
This commit, by only rescheduling the xmit timer once per ACK,
simplifies the code and reduces CPU overhead.
This commit was tested in an A/B test with Google web server
traffic. SNMP stats and request latency metrics were within noise
levels, substantiating that for normal web traffic patterns this is a
rare issue. This commit was also tested with packetdrill tests to
verify that it fixes the timer behavior in the corner cases discussed
in the netdev threads mentioned above.
This patch is a bug fix patch intended to be queued for -stable
relases.
Fixes: 6ba8a3b19e ("tcp: Tail loss probe (TLP)")
Reported-by: Klavs Klavsen <kl@vsen.dk>
Reported-by: Mao Wenan <maowenan@huawei.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Have tcp_schedule_loss_probe() base the TLP scheduling decision based
on when the RTO *should* fire. This is to enable the upcoming xmit
timer fix in this series, where tcp_schedule_loss_probe() cannot
assume that the last timer installed was an RTO timer (because we are
no longer doing the "rearm RTO, rearm RTO, rearm TLP" dance on every
ACK). So tcp_schedule_loss_probe() must independently figure out when
an RTO would want to fire.
In the new TLP implementation following in this series, we cannot
assume that icsk_timeout was set based on an RTO; after processing a
cumulative ACK the icsk_timeout we see can be from a previous TLP or
RTO. So we need to independently recalculate the RTO time (instead of
reading it out of icsk_timeout). Removing this dependency on the
nature of icsk_timeout makes things a little easier to reason about
anyway.
Note that the old and new code should be equivalent, since they are
both saying: "if the RTO is in the future, but at an earlier time than
the normal TLP time, then set the TLP timer to fire when the RTO would
have fired".
Fixes: 6ba8a3b19e ("tcp: Tail loss probe (TLP)")
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pure refactor. This helper will be required in the xmit timer fix
later in the patch series. (Because the TLP logic will want to make
this calculation.)
Fixes: 6ba8a3b19e ("tcp: Tail loss probe (TLP)")
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After commit c2ed1880fd ("net: ipv6: check route protocol when
deleting routes"), ipv6 route checks rt protocol when trying to
remove a rt entry.
It introduced a side effect causing 'ip -6 route flush cache' not
to work well. When flushing caches with iproute, all route caches
get dumped from kernel then removed one by one by sending DELROUTE
requests to kernel for each cache.
The thing is iproute sends the request with the cache whose proto
is set with RTPROT_REDIRECT by rt6_fill_node() when kernel dumps
it. But in kernel the rt_cache protocol is still 0, which causes
the cache not to be matched and removed.
So the real reason is rt6i_protocol in the route is not set when
it is allocated. As David Ahern's suggestion, this patch is to
set rt6i_protocol properly in the route when it is installed and
remove the codes setting rtm_protocol according to rt6i_flags in
rt6_fill_node.
This is also an improvement to keep rt6i_protocol consistent with
rtm_protocol.
Fixes: c2ed1880fd ("net: ipv6: check route protocol when deleting routes")
Reported-by: Jianlin Shi <jishi@redhat.com>
Suggested-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
- fix TT sync flag inconsistency problems, which can lead to excess packets,
by Linus Luessing
-----BEGIN PGP SIGNATURE-----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=a4HS
-----END PGP SIGNATURE-----
Merge tag 'batadv-net-for-davem-20170802' of git://git.open-mesh.org/linux-merge
Simon Wunderlich says:
====================
Here is a batman-adv bugfix:
- fix TT sync flag inconsistency problems, which can lead to excess packets,
by Linus Luessing
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
If the sender switches the congestion control during ECN-triggered
cwnd-reduction state (CA_CWR), upon exiting recovery cwnd is set to
the ssthresh value calculated by the previous congestion control. If
the previous congestion control is BBR that always keep ssthresh
to TCP_INIFINITE_SSTHRESH, cwnd ends up being infinite. The safe
step is to avoid assigning invalid ssthresh value when recovery ends.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the case that GRO is turned on and the original received packet is
CHECKSUM_PARTIAL, if the outer UDP header is exactly at the last
csum-unnecessary point, which for instance could occur if the packet
comes from another Linux guest on the same Linux host, we have to do
either remcsum_adjust or set up CHECKSUM_PARTIAL again with its
csum_start properly reset considering RCO.
However, since b7fe10e5eb ("gro: Fix remcsum offload to deal with frags
in GRO") that barrier in such case could be skipped if GRO turned on,
hence we pass over it and the inner L4 validation mistakenly reckons
it as a bad csum.
This patch makes remcsum_offload being reset at the same time of GRO
remcsum cleanup, so as to make it work in such case as before.
Fixes: b7fe10e5eb ("gro: Fix remcsum offload to deal with frags in GRO")
Signed-off-by: Koichiro Den <den@klaipeden.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
in for(),if((optlen > 0) && (optptr[1] == 0)), enter infinite loop.
Test: receive a packet which the ip length > 20 and the first byte of ip option is 0, produce this issue
Signed-off-by: yujuan.qi <yujuan.qi@mediatek.com>
Acked-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking fixes from David Miller:
1) Handle notifier registry failures properly in tun/tap driver, from
Tonghao Zhang.
2) Fix bpf verifier handling of subtraction bounds and add a testcase
for this, from Edward Cree.
3) Increase reset timeout in ftgmac100 driver, from Ben Herrenschmidt.
4) Fix use after free in prd_retire_rx_blk_timer_exired() in AF_PACKET,
from Cong Wang.
5) Fix SElinux regression due to recent UDP optimizations, from Paolo
Abeni.
6) We accidently increment IPSTATS_MIB_FRAGFAILS in the ipv6 code
paths, fix from Stefano Brivio.
7) Fix some mem leaks in dccp, from Xin Long.
8) Adjust MDIO_BUS kconfig deps to avoid build errors, from Arnd
Bergmann.
9) Mac address length check and buffer size fixes from Cong Wang.
10) Don't leak sockets in ipv6 udp early demux, from Paolo Abeni.
11) Fix return value when copy_from_user() fails in
bpf_prog_get_info_by_fd(), from Daniel Borkmann.
12) Handle PHY_HALTED properly in phy library state machine, from
Florian Fainelli.
13) Fix OOPS in fib_sync_down_dev(), from Ido Schimmel.
14) Fix truesize calculation in virtio_net which led to performance
regressions, from Michael S Tsirkin.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (76 commits)
samples/bpf: fix bpf tunnel cleanup
udp6: fix jumbogram reception
ppp: Fix a scheduling-while-atomic bug in del_chan
Revert "net: bcmgenet: Remove init parameter from bcmgenet_mii_config"
virtio_net: fix truesize for mergeable buffers
mv643xx_eth: fix of_irq_to_resource() error check
MAINTAINERS: Add more files to the PHY LIBRARY section
ipv4: fib: Fix NULL pointer deref during fib_sync_down_dev()
net: phy: Correctly process PHY_HALTED in phy_stop_machine()
sunhme: fix up GREG_STAT and GREG_IMASK register offsets
bpf: fix bpf_prog_get_info_by_fd to dump correct xlated_prog_len
tcp: avoid bogus gcc-7 array-bounds warning
net: tc35815: fix spelling mistake: "Intterrupt" -> "Interrupt"
bpf: don't indicate success when copy_from_user fails
udp6: fix socket leak on early demux
net: thunderx: Fix BGX transmit stall due to underflow
Revert "vhost: cache used event for better performance"
team: use a larger struct for mac address
net: check dev->addr_len for dev_set_mac_address()
phy: bcm-ns-usb3: fix MDIO_BUS dependency
...
Since commit 67a51780ae ("ipv6: udp: leverage scratch area
helpers") udp6_recvmsg() read the skb len from the scratch area,
to avoid a cache miss.
But the UDP6 rx path support RFC 2675 UDPv6 jumbograms, and their
length exceeds the 16 bits available in the scratch area. As a side
effect the length returned by recvmsg() is:
<ingress datagram len> % (1<<16)
This commit addresses the issue allocating one more bit in the
IP6CB flags field and setting it for incoming jumbograms.
Such field is still in the first cacheline, so at recvmsg()
time we can check it and fallback to access skb->len if
required, without a measurable overhead.
Fixes: 67a51780ae ("ipv6: udp: leverage scratch area helpers")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Michał reported a NULL pointer deref during fib_sync_down_dev() when
unregistering a netdevice. The problem is that we don't check for
'in_dev' being NULL, which can happen in very specific cases.
Usually routes are flushed upon NETDEV_DOWN sent in either the netdev or
the inetaddr notification chains. However, if an interface isn't
configured with any IP address, then it's possible for host routes to be
flushed following NETDEV_UNREGISTER, after NULLing dev->ip_ptr in
inetdev_destroy().
To reproduce:
$ ip link add type dummy
$ ip route add local 1.1.1.0/24 dev dummy0
$ ip link del dev dummy0
Fix this by checking for the presence of 'in_dev' before referencing it.
Fixes: 982acb9756 ("ipv4: fib: Notify about nexthop status changes")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reported-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Tested-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch fixes an issue in the translation table code potentially
leading to a TT Request + Response storm. The issue may occur for nodes
involving BLA and an inconsistent configuration of the batman-adv AP
isolation feature. However, since the new multicast optimizations, a
single, malformed packet may lead to a mesh-wide, persistent
Denial-of-Service, too.
The issue occurs because nodes are currently OR-ing the TT sync flags of
all originators announcing a specific MAC address via the
translation table. When an intermediate node now receives a TT Request
and wants to answer this on behalf of the destination node, then this
intermediate node now responds with an altered flag field and broken
CRC. The next OGM of the real destination will lead to a CRC mismatch
and triggering a TT Request and Response again.
Furthermore, the OR-ing is currently never undone as long as at least
one originator announcing the according MAC address remains, leading to
the potential persistency of this issue.
This patch fixes this issue by storing the flags used in the CRC
calculation on a a per TT orig entry basis to be able to respond with
the correct, original flags in an intermediate TT Response for one
thing. And to be able to correctly unset sync flags once all nodes
announcing a sync flag vanish for another.
Fixes: e9c00136a4 ("batman-adv: fix tt_global_entries flags update")
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Acked-by: Antonio Quartulli <a@unstable.cc>
[sw: typo in commit message]
Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
When using CONFIG_UBSAN_SANITIZE_ALL, the TCP code produces a
false-positive warning:
net/ipv4/tcp_output.c: In function 'tcp_connect':
net/ipv4/tcp_output.c:2207:40: error: array subscript is below array bounds [-Werror=array-bounds]
tp->chrono_stat[tp->chrono_type - 1] += now - tp->chrono_start;
^~
net/ipv4/tcp_output.c:2207:40: error: array subscript is below array bounds [-Werror=array-bounds]
tp->chrono_stat[tp->chrono_type - 1] += now - tp->chrono_start;
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~
I have opened a gcc bug for this, but distros have already shipped
compilers with this problem, and it's not clear yet whether there is
a way for gcc to avoid the warning. As the problem is related to the
bitfield access, this introduces a temporary variable to store the old
enum value.
I did not notice this warning earlier, since UBSAN is disabled when
building with COMPILE_TEST, and that was always turned on in both
allmodconfig and randconfig tests.
Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81601
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
When an early demuxed packet reaches __udp6_lib_lookup_skb(), the
sk reference is retrieved and used, but the relevant reference
count is leaked and the socket destructor is never called.
Beyond leaking the sk memory, if there are pending UDP packets
in the receive queue, even the related accounted memory is leaked.
In the long run, this will cause persistent forward allocation errors
and no UDP skbs (both ipv4 and ipv6) will be able to reach the
user-space.
Fix this by explicitly accessing the early demux reference before
the lookup, and properly decreasing the socket reference count
after usage.
Also drop the skb_steal_sock() in __udp6_lib_lookup_skb(), and
the now obsoleted comment about "socket cache".
The newly added code is derived from the current ipv4 code for the
similar path.
v1 -> v2:
fixed the __udp6_lib_rcv() return code for resubmission,
as suggested by Eric
Reported-by: Sam Edwards <CFSworks@gmail.com>
Reported-by: Marc Haber <mh+netdev@zugschlus.de>
Fixes: 5425077d73 ("net: ipv6: Add early demux handler for UDP unicast")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Historically, dev_ifsioc() uses struct sockaddr as mac
address definition, this is why dev_set_mac_address()
accepts a struct sockaddr pointer as input but now we
have various types of mac addresse whose lengths
are up to MAX_ADDR_LEN, longer than struct sockaddr,
and saved in dev->addr_len.
It is too late to fix dev_ifsioc() due to API
compatibility, so just reject those larger than
sizeof(struct sockaddr), otherwise we would read
and use some random bytes from kernel stack.
Fortunately, only a few IPv6 tunnel devices have addr_len
larger than sizeof(struct sockaddr) and they don't support
ndo_set_mac_addr(). But with team driver, in lb mode, they
can still be enslaved to a team master and make its mac addr
length as the same.
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In dccp_feat_init, when ccid_get_builtin_ccids failsto alloc
memory for rx.val, it should free tx.val before returning an
error.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The patch "dccp: fix a memleak that dccp_ipv6 doesn't put reqsk
properly" fixed reqsk refcnt leak for dccp_ipv6. The same issue
exists on dccp_ipv4.
This patch is to fix it for dccp_ipv4.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In dccp_v6_conn_request, after reqsk gets alloced and hashed into
ehash table, reqsk's refcnt is set 3. one is for req->rsk_timer,
one is for hlist, and the other one is for current using.
The problem is when dccp_v6_conn_request returns and finishes using
reqsk, it doesn't put reqsk. This will cause reqsk refcnt leaks and
reqsk obj never gets freed.
Jianlin found this issue when running dccp_memleak.c in a loop, the
system memory would run out.
dccp_memleak.c:
int s1 = socket(PF_INET6, 6, IPPROTO_IP);
bind(s1, &sa1, 0x20);
listen(s1, 0x9);
int s2 = socket(PF_INET6, 6, IPPROTO_IP);
connect(s2, &sa1, 0x20);
close(s1);
close(s2);
This patch is to put the reqsk before dccp_v6_conn_request returns,
just as what tcp_conn_request does.
Reported-by: Jianlin Shi <jishi@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Apparently netpoll_setup() assumes that netpoll.dev_name is a pointer
when checking if the device name is set:
if (np->dev_name) {
...
However the field is a character array, therefore the condition always
yields true. Check instead whether the first byte of the array has a
non-zero value.
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
We must use pre-processor conditional block or suitable accessors to
manipulate skb->sp elsewhere builds lacking the CONFIG_XFRM will break.
Fixes: dce4551cb2 ("udp: preserve head state for IP_CMSG_PASSSEC")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
RFC 2465 defines ipv6IfStatsOutFragFails as:
"The number of IPv6 datagrams that have been discarded
because they needed to be fragmented at this output
interface but could not be."
The existing implementation, instead, would increase the counter
twice in case we fail to allocate room for single fragments:
once for the fragment, once for the datagram.
This didn't look intentional though. In one of the two affected
affected failure paths, the double increase was simply a result
of a new 'goto fail' statement, introduced to avoid a skb leak.
The other path appears to be affected since at least 2.6.12-rc2.
Reported-by: Sabrina Dubroca <sdubroca@redhat.com>
Fixes: 1d325d217c ("ipv6: ip6_fragment: fix headroom tests and skb leak")
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Paul Moore reported a SELinux/IP_PASSSEC regression
caused by missing skb->sp at recvmsg() time. We need to
preserve the skb head state to process the IP_CMSG_PASSSEC
cmsg.
With this commit we avoid releasing the skb head state in the
BH even if a secpath is attached to the current skb, and stores
the skb status (with/without head states) in the scratch area,
so that we can access it at skb deallocation time, without
incurring in cache-miss penalties.
This also avoids misusing the skb CB for ipv6 packets,
as introduced by the commit 0ddf3fb2c4 ("udp: preserve
skb->dst if required for IP options processing").
Clean a bit the scratch area helpers implementation, to
reduce the code differences between 32 and 64 bits build.
Reported-by: Paul Moore <paul@paul-moore.com>
Fixes: 0a463c78d2 ("udp: avoid a cache miss on dequeue")
Fixes: 0ddf3fb2c4 ("udp: preserve skb->dst if required for IP options processing")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Tested-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mt7530 driver has its dsa_switch_ops::get_tag_protocol function
check ds->cpu_port_mask to issue a warning in case the configured CPU
port is not capable of supporting tags.
After commit 14be36c2c9 ("net: dsa: Initialize all CPU and enabled
ports masks in dsa_ds_parse()") we slightly re-arranged the
initialization such that this was no longer working. Just make sure that
ds->cpu_port_mask is set prior to the first call to get_tag_protocol,
thus restoring the expected contract. In case of error, the CPU port bit
is cleared.
Fixes: 14be36c2c9 ("net: dsa: Initialize all CPU and enabled ports masks in dsa_ds_parse()")
Reported-by: Sean Wang <sean.wang@mediatek.com>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are multiple reports showing we have a use-after-free in
the timer prb_retire_rx_blk_timer_expired(), where we use struct
tpacket_kbdq_core::pkbdq, a pg_vec, after it gets freed by
free_pg_vec().
The interesting part is it is not freed via packet_release() but
via packet_setsockopt(), which means we are not closing the socket.
Looking into the big and fat function packet_set_ring(), this could
happen if we satisfy the following conditions:
1. closing == 0, not on packet_release() path
2. req->tp_block_nr == 0, we don't allocate a new pg_vec
3. rx_ring->pg_vec is already set as V3, which means we already called
packet_set_ring() wtih req->tp_block_nr > 0 previously
4. req->tp_frame_nr == 0, pass sanity check
5. po->mapped == 0, never called mmap()
In this scenario we are clearing the old rx_ring->pg_vec, so we need
to free this pg_vec, but we don't stop the timer on this path because
of closing==0.
The timer has to be stopped as long as we need to free pg_vec, therefore
the check on closing!=0 is wrong, we should check pg_vec!=NULL instead.
Thanks to liujian for testing different fixes.
Reported-by: alexander.levin@verizon.com
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Reported-by: liujian (CE) <liujian56@huawei.com>
Tested-by: liujian (CE) <liujian56@huawei.com>
Cc: Ding Tianhong <dingtianhong@huawei.com>
Cc: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Before the 'type' is validated, we shouldn't use it to fetch the
ovs_ct_attr_lens's minlen and maxlen, else, out of bound access
may happen.
Fixes: 7f8a436eaa ("openvswitch: Add conntrack action")
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The commit ffb07550c7 ("copy_msghdr_from_user(): get rid of
field-by-field copyin") introduce a new sparse warning:
net/socket.c:1919:27: warning: incorrect type in assignment (different address spaces)
net/socket.c:1919:27: expected void *msg_control
net/socket.c:1919:27: got void [noderef] <asn:1>*[addressable] msg_control
and a line above 80 chars, let's fix them
Fixes: ffb07550c7 ("copy_msghdr_from_user(): get rid of field-by-field copyin")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 3d4762639d ("tcp: remove poll() flakes when receiving
RST") in v4.12 changed the order in which ->sk_state_change()
and ->sk_error_report() are called when a socket is shut
down - sk_state_change() is now called first.
This causes xs_tcp_state_change() -> xs_sock_mark_closed() ->
xprt_disconnect_done() to wake all pending tasked with -EAGAIN.
When the ->sk_error_report() callback arrives, it is too late to
pass the error on, and it is lost.
As easy way to demonstrate the problem caused is to try to start
rpc.nfsd while rcpbind isn't running.
nfsd will attempt a tcp connection to rpcbind. A ECONNREFUSED
error is returned, but sunrpc code loses the error and keeps
retrying. If it saw the ECONNREFUSED, it would abort.
To fix this, handle the sk->sk_err in the TCP_CLOSE branch of
xs_tcp_state_change().
Fixes: 3d4762639d ("tcp: remove poll() flakes when receiving RST")
Cc: stable@vger.kernel.org (v4.12)
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Pull networking fixes from David Miller:
1) BPF verifier signed/unsigned value tracking fix, from Daniel
Borkmann, Edward Cree, and Josef Bacik.
2) Fix memory allocation length when setting up calls to
->ndo_set_mac_address, from Cong Wang.
3) Add a new cxgb4 device ID, from Ganesh Goudar.
4) Fix FIB refcount handling, we have to set it's initial value before
the configure callback (which can bump it). From David Ahern.
5) Fix double-free in qcom/emac driver, from Timur Tabi.
6) A bunch of gcc-7 string format overflow warning fixes from Arnd
Bergmann.
7) Fix link level headroom tests in ip_do_fragment(), from Vasily
Averin.
8) Fix chunk walking in SCTP when iterating over error and parameter
headers. From Alexander Potapenko.
9) TCP BBR congestion control fixes from Neal Cardwell.
10) Fix SKB fragment handling in bcmgenet driver, from Doug Berger.
11) BPF_CGROUP_RUN_PROG_SOCK_OPS needs to check for null __sk, from Cong
Wang.
12) xmit_recursion in ppp driver needs to be per-device not per-cpu,
from Gao Feng.
13) Cannot release skb->dst in UDP if IP options processing needs it.
From Paolo Abeni.
14) Some netdev ioctl ifr_name[] NULL termination fixes. From Alexander
Levin and myself.
15) Revert some rtnetlink notification changes that are causing
regressions, from David Ahern.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (83 commits)
net: bonding: Fix transmit load balancing in balance-alb mode
rds: Make sure updates to cp_send_gen can be observed
net: ethernet: ti: cpsw: Push the request_irq function to the end of probe
ipv4: initialize fib_trie prior to register_netdev_notifier call.
rtnetlink: allocate more memory for dev_set_mac_address()
net: dsa: b53: Add missing ARL entries for BCM53125
bpf: more tests for mixed signed and unsigned bounds checks
bpf: add test for mixed signed and unsigned bounds checks
bpf: fix up test cases with mixed signed/unsigned bounds
bpf: allow to specify log level and reduce it for test_verifier
bpf: fix mixed signed/unsigned derived min/max value bounds
ipv6: avoid overflow of offset in ip6_find_1stfragopt
net: tehuti: don't process data if it has not been copied from userspace
Revert "rtnetlink: Do not generate notifications for CHANGEADDR event"
net: dsa: mv88e6xxx: Enable CMODE config support for 6390X
dt-binding: ptp: Add SoC compatibility strings for dte ptp clock
NET: dwmac: Make dwmac reset unconditional
net: Zero terminate ifr_name in dev_ifname().
wireless: wext: terminate ifr name coming from userspace
netfilter: fix netfilter_net_init() return
...
cp->cp_send_gen is treated as a normal variable, although it may be
used by different threads.
This is fixed by using {READ,WRITE}_ONCE when it is incremented and
READ_ONCE when it is read outside the {acquire,release}_in_xmit
protection.
Normative reference from the Linux-Kernel Memory Model:
Loads from and stores to shared (but non-atomic) variables should
be protected with the READ_ONCE(), WRITE_ONCE(), and
ACCESS_ONCE().
Clause 5.1.2.4/25 in the C standard is also relevant.
Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Reviewed-by: Knut Omang <knut.omang@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Net stack initialization currently initializes fib-trie after the
first call to netdevice_notifier() call. In fact fib_trie initialization
needs to happen before first rtnl_register(). It does not cause any problem
since there are no devices UP at this moment, but trying to bring 'lo'
UP at initialization would make this assumption wrong and exposes the issue.
Fixes following crash
Call Trace:
? alternate_node_alloc+0x76/0xa0
fib_table_insert+0x1b7/0x4b0
fib_magic.isra.17+0xea/0x120
fib_add_ifaddr+0x7b/0x190
fib_netdev_event+0xc0/0x130
register_netdevice_notifier+0x1c1/0x1d0
ip_fib_init+0x72/0x85
ip_rt_init+0x187/0x1e9
ip_init+0xe/0x1a
inet_init+0x171/0x26c
? ipv4_offload_init+0x66/0x66
do_one_initcall+0x43/0x160
kernel_init_freeable+0x191/0x219
? rest_init+0x80/0x80
kernel_init+0xe/0x150
ret_from_fork+0x22/0x30
Code: f6 46 23 04 74 86 4c 89 f7 e8 ae 45 01 00 49 89 c7 4d 85 ff 0f 85 7b ff ff ff 31 db eb 08 4c 89 ff e8 16 47 01 00 48 8b 44 24 38 <45> 8b 6e 14 4d 63 76 74 48 89 04 24 0f 1f 44 00 00 48 83 c4 08
RIP: kmem_cache_alloc+0xcf/0x1c0 RSP: ffff9b1500017c28
CR2: 0000000000000014
Fixes: 7b1a74fdbb ("[NETNS]: Refactor fib initialization so it can handle multiple namespaces.")
Fixes: 7f9b80529b ("[IPV4]: fib hash|trie initialization")
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
virtnet_set_mac_address() interprets mac address as struct
sockaddr, but upper layer only allocates dev->addr_len
which is ETH_ALEN + sizeof(sa_family_t) in this case.
We lack a unified definition for mac address, so just fix
the upper layer, this also allows drivers to interpret it
to struct sockaddr freely.
Reported-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In some cases, offset can overflow and can cause an infinite loop in
ip6_find_1stfragopt(). Make it unsigned int to prevent the overflow, and
cap it at IPV6_MAXPLEN, since packets larger than that should be invalid.
This problem has been here since before the beginning of git history.
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit cd8966e75e.
The duplicate CHANGEADDR event message is sent regardless of link
status whereas the setlink changes only generate a notification when
the link is up. Not sending a notification when the link is down breaks
dhcpcd which only processes hwaddr changes when the link is down.
Fixes reported regression:
https://bugzilla.kernel.org/show_bug.cgi?id=196355
Reported-by: Yaroslav Isakov <yaroslav.isakov@gmail.com>
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ifr name is assumed to be a valid string by the kernel, but nothing
was forcing username to pass a valid string.
In turn, this would cause panics as we tried to access the string
past it's valid memory.
Signed-off-by: Sasha Levin <alexander.levin@verizon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We accidentally return an uninitialized variable.
Fixes: cf56c2f892 ("netfilter: remove old pre-netns era hook api")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pablo Neira Ayuso says:
====================
Netfilter fixes for net
The following patchset contains Netfilter fixes for your net tree,
they are:
1) Missing netlink message sanity check in nfnetlink, patch from
Mateusz Jurczyk.
2) We now have netfilter per-netns hooks, so let's kill global hook
infrastructure, this infrastructure is known to be racy with netns.
We don't care about out of tree modules. Patch from Florian Westphal.
3) find_appropriate_src() is buggy when colissions happens after the
conversion of the nat bysource to rhashtable. Also from Florian.
4) Remove forward chain in nf_tables arp family, it's useless and it is
causing quite a bit of confusion, from Florian Westphal.
5) nf_ct_remove_expect() is called with the wrong parameter, causing
kernel oops, patch from Florian Westphal.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric noticed that in udp_recvmsg() we still need to access
skb->dst while processing the IP options.
Since commit 0a463c78d2 ("udp: avoid a cache miss on dequeue")
skb->dst is no more available at recvmsg() time and bad things
will happen if we enter the relevant code path.
This commit address the issue, avoid clearing skb->dst if
any IP options are present into the relevant skb.
Since the IP CB is contained in the first skb cacheline, we can
test it to decide to leverage the consume_stateless_skb()
optimization, without measurable additional cost in the faster
path.
v1 -> v2: updated commit message tags
Fixes: 0a463c78d2 ("udp: avoid a cache miss on dequeue")
Reported-by: Andrey Konovalov <andreyknvl@google.com>
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We crash in __nf_ct_expect_check, it calls nf_ct_remove_expect on the
uninitialised expectation instead of existing one, so del_timer chokes
on random memory address.
Fixes: ec0e3f0111 ("netfilter: nf_ct_expect: Add nf_ct_remove_expect()")
Reported-by: Sergey Kvachonok <ravenexp@gmail.com>
Tested-by: Sergey Kvachonok <ravenexp@gmail.com>
Cc: Gao Feng <fgao@ikuai8.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
arp packets cannot be forwarded.
They can be bridged, but then they can be filtered using
either ebtables or nftables bridge family.
The bridge netfilter exposes a "call-arptables" switch which
pushes packets into arptables, but lets not expose this for nftables, so better
close this asap.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
When doing initial conversion to rhashtable I replaced the bucket
walk with a single rhashtable_lookup_fast().
When moving to rhlist I failed to properly walk the list of identical
tuples, but that is what is needed for this to work correctly.
The table contains the original tuples, so the reply tuples are all
distinct.
We currently decide that mapping is (not) in range only based on the
first entry, but in case its not we need to try the reply tuple of the
next entry until we either find an in-range mapping or we checked
all the entries.
This bug makes nat core attempt collision resolution while it might be
able to use the mapping as-is.
Fixes: 870190a9ec ("netfilter: nat: convert nat bysrc hash to rhashtable")
Reported-by: Jaco Kroon <jaco@uls.co.za>
Tested-by: Jaco Kroon <jaco@uls.co.za>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
no more users in the tree, remove this.
The old api is racy wrt. module removal, all users have been converted
to the netns-aware api.
The old api pretended we still have global hooks but that has not been
true for a long time.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
If kmem_cache_zalloc() returns NULL then the INIT_LIST_HEAD(&data->links);
will Oops. The callers aren't really prepared for NULL returns so it
doesn't make a lot of difference in real life.
Fixes: 5240d9f95d ("libceph: replace message data pointer with list")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
... otherwise we die in insert_pg_mapping(), which wants pg->node to be
empty, i.e. initialized with RB_CLEAR_NODE.
Fixes: 6f428df47d ("libceph: pg_upmap[_items] infrastructure")
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
No sooner than Dan had fixed this issue in commit 293dffaad8
("libceph: NULL deref on crush_decode() error path"), I brought it
back. Add a new label and set -EINVAL once, right before failing.
Fixes: 278b1d709c ("libceph: ceph_decode_skip_* helpers")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
There are hidden gotos in the ceph_decode_* macros. We need to set the
"err" variable on these error paths otherwise we end up returning
ERR_PTR(0) which is NULL. It causes NULL dereferences in the callers.
Fixes: 6f428df47d ("libceph: pg_upmap[_items] infrastructure")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
[idryomov@gmail.com: similar bug in osdmap_decode(), changelog tweak]
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Neal Cardwell