We've been getting checksum errors involving small UDP packets, usually
59B packets with 1 extra non-zero padding byte. netdev_rx_csum_fault()
has been complaining that HW is providing bad checksums. Turns out the
problem is in pskb_trim_rcsum_slow(), introduced in commit 88078d98d1
("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends").
The source of the problem is that when the bytes we are trimming start
at an odd address, as in the case of the 1 padding byte above,
skb_checksum() returns a byte-swapped value. We cannot just combine this
with skb->csum using csum_sub(). We need to use csum_block_sub() here
that takes into account the parity of the start address and handles the
swapping.
Matches existing code in __skb_postpull_rcsum() and esp_remove_trailer().
Fixes: 88078d98d1 ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
Signed-off-by: Dimitris Michailidis <dmichail@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit 6fe9487892.
It is causing more serious regressions than the RCU warning
it is fixing.
Signed-off-by: David S. Miller <davem@davemloft.net>
In dev_ethtool(), the eth command 'ethcmd' is firstly copied from the
use-space buffer 'useraddr' and checked to see whether it is
ETHTOOL_PERQUEUE. If yes, the sub-command 'sub_cmd' is further copied from
the user space. Otherwise, 'sub_cmd' is the same as 'ethcmd'. Next,
according to 'sub_cmd', a permission check is enforced through the function
ns_capable(). For example, the permission check is required if 'sub_cmd' is
ETHTOOL_SCOALESCE, but it is not necessary if 'sub_cmd' is
ETHTOOL_GCOALESCE, as suggested in the comment "Allow some commands to be
done by anyone". The following execution invokes different handlers
according to 'ethcmd'. Specifically, if 'ethcmd' is ETHTOOL_PERQUEUE,
ethtool_set_per_queue() is called. In ethtool_set_per_queue(), the kernel
object 'per_queue_opt' is copied again from the user-space buffer
'useraddr' and 'per_queue_opt.sub_command' is used to determine which
operation should be performed. Given that the buffer 'useraddr' is in the
user space, a malicious user can race to change the sub-command between the
two copies. In particular, the attacker can supply ETHTOOL_PERQUEUE and
ETHTOOL_GCOALESCE to bypass the permission check in dev_ethtool(). Then
before ethtool_set_per_queue() is called, the attacker changes
ETHTOOL_GCOALESCE to ETHTOOL_SCOALESCE. In this way, the attacker can
bypass the permission check and execute ETHTOOL_SCOALESCE.
This patch enforces a check in ethtool_set_per_queue() after the second
copy from 'useraddr'. If the sub-command is different from the one obtained
in the first copy in dev_ethtool(), an error code EINVAL will be returned.
Fixes: f38d138a7d ("net/ethtool: support set coalesce per queue")
Signed-off-by: Wenwen Wang <wang6495@umn.edu>
Reviewed-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
In ethtool_get_rxnfc(), the eth command 'cmd' is compared against
'ETHTOOL_GRXFH' to see whether it is necessary to adjust the variable
'info_size'. Then the whole structure of 'info' is copied from the
user-space buffer 'useraddr' with 'info_size' bytes. In the following
execution, 'info' may be copied again from the buffer 'useraddr' depending
on the 'cmd' and the 'info.flow_type'. However, after these two copies,
there is no check between 'cmd' and 'info.cmd'. In fact, 'cmd' is also
copied from the buffer 'useraddr' in dev_ethtool(), which is the caller
function of ethtool_get_rxnfc(). Given that 'useraddr' is in the user
space, a malicious user can race to change the eth command in the buffer
between these copies. By doing so, the attacker can supply inconsistent
data and cause undefined behavior because in the following execution 'info'
will be passed to ops->get_rxnfc().
This patch adds a necessary check on 'info.cmd' and 'cmd' to confirm that
they are still same after the two copies in ethtool_get_rxnfc(). Otherwise,
an error code EINVAL will be returned.
Signed-off-by: Wenwen Wang <wang6495@umn.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since commit 5aad1de5ea ("ipv4: use separate genid for next hop
exceptions"), exceptions get deprecated separately from cached
routes. In particular, administrative changes don't clear PMTU anymore.
As Stefano described in commit e9fa1495d7 ("ipv6: Reflect MTU changes
on PMTU of exceptions for MTU-less routes"), the PMTU discovered before
the local MTU change can become stale:
- if the local MTU is now lower than the PMTU, that PMTU is now
incorrect
- if the local MTU was the lowest value in the path, and is increased,
we might discover a higher PMTU
Similarly to what commit e9fa1495d7 did for IPv6, update PMTU in those
cases.
If the exception was locked, the discovered PMTU was smaller than the
minimal accepted PMTU. In that case, if the new local MTU is smaller
than the current PMTU, let PMTU discovery figure out if locking of the
exception is still needed.
To do this, we need to know the old link MTU in the NETDEV_CHANGEMTU
notifier. By the time the notifier is called, dev->mtu has been
changed. This patch adds the old MTU as additional information in the
notifier structure, and a new call_netdevice_notifiers_u32() function.
Fixes: 5aad1de5ea ("ipv4: use separate genid for next hop exceptions")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Devlink string param buffer is allocated at the size of
DEVLINK_PARAM_MAX_STRING_VALUE. Add helper function which makes sure
this size is not exceeded.
Renamed DEVLINK_PARAM_MAX_STRING_VALUE to
__DEVLINK_PARAM_MAX_STRING_VALUE to emphasize that it should be used by
devlink only. The driver should use the helper function instead to
verify it doesn't exceed the allowed length.
Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Driverinit configuration mode value is held by devlink to enable the
driver fetch the value after reload command. In case the param type is
string devlink should copy the value from driver string buffer to
devlink string buffer on devlink_param_driverinit_value_set() and
vice-versa on devlink_param_driverinit_value_get().
Fixes: ec01aeb180 ("devlink: Add support for get/set driverinit value")
Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In case devlink param type is string, it needs to copy the string value
it got from the input to devlink_param_value.
Fixes: e3b7ca18ad ("devlink: Add param set command")
Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, rtnl_fdb_dump() assumes the family header is 'struct ifinfomsg',
which is not always true -- 'struct ndmsg' is used by iproute2 ('ip neigh').
The problem is, the function bails out early if nlmsg_parse() fails, which
does occur for iproute2 usage of 'struct ndmsg' because the payload length
is shorter than the family header alone (as 'struct ifinfomsg' is assumed).
This breaks backward compatibility with userspace -- nothing is sent back.
Some examples with iproute2 and netlink library for go [1]:
1) $ bridge fdb show
33:33:00:00:00:01 dev ens3 self permanent
01:00:5e:00:00:01 dev ens3 self permanent
33:33:ff:15:98:30 dev ens3 self permanent
This one works, as it uses 'struct ifinfomsg'.
fdb_show() @ iproute2/bridge/fdb.c
"""
.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg)),
...
if (rtnl_dump_request(&rth, RTM_GETNEIGH, [...]
"""
2) $ ip --family bridge neigh
RTNETLINK answers: Invalid argument
Dump terminated
This one fails, as it uses 'struct ndmsg'.
do_show_or_flush() @ iproute2/ip/ipneigh.c
"""
.n.nlmsg_type = RTM_GETNEIGH,
.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)),
"""
3) $ ./neighlist
< no output >
This one fails, as it uses 'struct ndmsg'-based.
neighList() @ netlink/neigh_linux.go
"""
req := h.newNetlinkRequest(unix.RTM_GETNEIGH, [...]
msg := Ndmsg{
"""
The actual breakage was introduced by commit 0ff50e83b5 ("net: rtnetlink:
bail out from rtnl_fdb_dump() on parse error"), because nlmsg_parse() fails
if the payload length (with the _actual_ family header) is less than the
family header length alone (which is assumed, in parameter 'hdrlen').
This is true in the examples above with struct ndmsg, with size and payload
length shorter than struct ifinfomsg.
However, that commit just intends to fix something under the assumption the
family header is indeed an 'struct ifinfomsg' - by preventing access to the
payload as such (via 'ifm' pointer) if the payload length is not sufficient
to actually contain it.
The assumption was introduced by commit 5e6d243587 ("bridge: netlink dump
interface at par with brctl"), to support iproute2's 'bridge fdb' command
(not 'ip neigh') which indeed uses 'struct ifinfomsg', thus is not broken.
So, in order to unbreak the 'struct ndmsg' family headers and still allow
'struct ifinfomsg' to continue to work, check for the known message sizes
used with 'struct ndmsg' in iproute2 (with zero or one attribute which is
not used in this function anyway) then do not parse the data as ifinfomsg.
Same examples with this patch applied (or revert/before the original fix):
$ bridge fdb show
33:33:00:00:00:01 dev ens3 self permanent
01:00:5e:00:00:01 dev ens3 self permanent
33:33:ff:15:98:30 dev ens3 self permanent
$ ip --family bridge neigh
dev ens3 lladdr 33:33:00:00:00:01 PERMANENT
dev ens3 lladdr 01:00:5e:00:00:01 PERMANENT
dev ens3 lladdr 33:33:ff:15:98:30 PERMANENT
$ ./neighlist
netlink.Neigh{LinkIndex:2, Family:7, State:128, Type:0, Flags:2, IP:net.IP(nil), HardwareAddr:net.HardwareAddr{0x33, 0x33, 0x0, 0x0, 0x0, 0x1}, LLIPAddr:net.IP(nil), Vlan:0, VNI:0}
netlink.Neigh{LinkIndex:2, Family:7, State:128, Type:0, Flags:2, IP:net.IP(nil), HardwareAddr:net.HardwareAddr{0x1, 0x0, 0x5e, 0x0, 0x0, 0x1}, LLIPAddr:net.IP(nil), Vlan:0, VNI:0}
netlink.Neigh{LinkIndex:2, Family:7, State:128, Type:0, Flags:2, IP:net.IP(nil), HardwareAddr:net.HardwareAddr{0x33, 0x33, 0xff, 0x15, 0x98, 0x30}, LLIPAddr:net.IP(nil), Vlan:0, VNI:0}
Tested on mainline (v4.19-rc6) and net-next (3bd09b05b0).
References:
[1] netlink library for go (test-case)
https://github.com/vishvananda/netlink
$ cat ~/go/src/neighlist/main.go
package main
import ("fmt"; "syscall"; "github.com/vishvananda/netlink")
func main() {
neighs, _ := netlink.NeighList(0, syscall.AF_BRIDGE)
for _, neigh := range neighs { fmt.Printf("%#v\n", neigh) }
}
$ export GOPATH=~/go
$ go get github.com/vishvananda/netlink
$ go build neighlist
$ ~/go/src/neighlist/neighlist
Thanks to David Ahern for suggestions to improve this patch.
Fixes: 0ff50e83b5 ("net: rtnetlink: bail out from rtnl_fdb_dump() on parse error")
Fixes: 5e6d243587 ("bridge: netlink dump interface at par with brctl")
Reported-by: Aidan Obley <aobley@pivotal.io>
Signed-off-by: Mauricio Faria de Oliveira <mfo@canonical.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We have an impressive number of syzkaller bugs that are linked
to the fact that syzbot was able to create a networking device
with millions of TX (or RX) queues.
Let's limit the number of RX/TX queues to 4096, this really should
cover all known cases.
A separate patch will add various cond_resched() in the loops
handling sysfs entries at device creation and dismantle.
Tested:
lpaa6:~# ip link add gre-4097 numtxqueues 4097 numrxqueues 4097 type ip6gretap
RTNETLINK answers: Invalid argument
lpaa6:~# time ip link add gre-4096 numtxqueues 4096 numrxqueues 4096 type ip6gretap
real 0m0.180s
user 0m0.000s
sys 0m0.107s
Fixes: 76ff5cc919 ("rtnl: allow to specify number of rx and tx queues on device creation")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Link dumps can return results from a target namespace. If the namespace id
is invalid, then the dump request should fail if get_target_net fails
rather than continuing with a dump of the current namespace.
Fixes: 79e1ad148c ("rtnetlink: use netnsid to query interface")
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since we do no longer require NAPI drivers to provide
an ndo_poll_controller(), napi_schedule() has not been done
before poll_one_napi() invocation.
So testing NAPI_STATE_SCHED is likely to cause early returns.
While we are at it, remove outdated comment.
Note to future bisections : This change might surface prior
bugs in drivers. See commit 73f21c653f ("bnxt_en: Fix TX
timeout during netpoll.") for one occurrence.
Fixes: ac3d9dd034 ("netpoll: make ndo_poll_controller() optional")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Tested-by: Song Liu <songliubraving@fb.com>
Cc: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add flag wol_enabled to struct net_device indicating whether
Wake-on-LAN is enabled. As first user phy_suspend() will use it to
decide whether PHY can be suspended or not.
Fixes: f1e911d5d0 ("r8169: add basic phylib support")
Fixes: e8cfd9d6c7 ("net: phy: call state machine synchronously in phy_stop")
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As diagnosed by Song Liu, ndo_poll_controller() can
be very dangerous on loaded hosts, since the cpu
calling ndo_poll_controller() might steal all NAPI
contexts (for all RX/TX queues of the NIC). This capture
can last for unlimited amount of time, since one
cpu is generally not able to drain all the queues under load.
It seems that all networking drivers that do use NAPI
for their TX completions, should not provide a ndo_poll_controller().
NAPI drivers have netpoll support already handled
in core networking stack, since netpoll_poll_dev()
uses poll_napi(dev) to iterate through registered
NAPI contexts for a device.
This patch allows netpoll_poll_dev() to process NAPI
contexts even for drivers not providing ndo_poll_controller(),
allowing for following patches in NAPI drivers.
Also we export netpoll_poll_dev() so that it can be called
by bonding/team drivers in following patches.
Reported-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Tested-by: Song Liu <songliubraving@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
So it should not fail with EPERM even though it is no longer implemented...
This is a fix for:
(userns)$ egrep ^Cap /proc/self/status
CapInh: 0000003fffffffff
CapPrm: 0000003fffffffff
CapEff: 0000003fffffffff
CapBnd: 0000003fffffffff
CapAmb: 0000003fffffffff
(userns)$ tcpdump -i usb_rndis0
tcpdump: WARNING: usb_rndis0: SIOCETHTOOL(ETHTOOL_GUFO) ioctl failed: Operation not permitted
Warning: Kernel filter failed: Bad file descriptor
tcpdump: can't remove kernel filter: Bad file descriptor
With this change it returns EOPNOTSUPP instead of EPERM.
See also https://github.com/the-tcpdump-group/libpcap/issues/689
Fixes: 08a00fea6d "net: Remove references to NETIF_F_UFO from ethtool."
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Smatch reports that devlink_dpipe_send_and_alloc_skb() frees the skb
on error so this is a double free. We fixed a bunch of these bugs in
commit 7fe4d6dcbc ("devlink: Remove redundant free on error path") but
we accidentally overlooked this one.
Fixes: d9f9b9a4d0 ("devlink: Add support for resource abstraction")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann says:
====================
pull-request: bpf 2018-09-16
The following pull-request contains BPF updates for your *net* tree.
The main changes are:
1) Fix end boundary calculation in BTF for the type section, from Martin.
2) Fix and revert subtraction of pointers that was accidentally allowed
for unprivileged programs, from Alexei.
3) Fix bpf_msg_pull_data() helper by using __GFP_COMP in order to avoid
a warning in linearizing sg pages into a single one for large allocs,
from Tushar.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Update 'confirmed' timestamp when ARP packet is received. It shouldn't
affect locktime logic and anyway entry can be confirmed by any higher-layer
protocol. Thus it makes sense to confirm it when ARP packet is received.
Fixes: 77d7123342 ("neighbour: update neigh timestamps iff update is effective")
Signed-off-by: Vasily Khoruzhick <vasilykh@arista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fix addresses https://bugzilla.kernel.org/show_bug.cgi?id=201071
Commit 5025f7f7d5 wrongly relied on __dev_change_flags to notify users of
dev flag changes in the case when dev->rtnl_link_state = RTNL_LINK_INITIALIZED.
Fix it by indicating flag changes explicitly to __dev_notify_flags.
Fixes: 5025f7f7d5 ("rtnetlink: add rtnl_link_state check in rtnl_configure_link")
Reported-By: Liam mcbirnie <liam.mcbirnie@boeing.com>
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Helper bpg_msg_pull_data() can allocate multiple pages while
linearizing multiple scatterlist elements into one shared page.
However, if the shared page has size > PAGE_SIZE, using
copy_page_to_iter() causes below warning.
e.g.
[ 6367.019832] WARNING: CPU: 2 PID: 7410 at lib/iov_iter.c:825
page_copy_sane.part.8+0x0/0x8
To avoid above warning, use __GFP_COMP while allocating multiple
contiguous pages.
Fixes: 015632bb30 ("bpf: sk_msg program helper bpf_sk_msg_pull_data")
Signed-off-by: Tushar Dave <tushar.n.dave@oracle.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
According to the documentation in msg_zerocopy.rst, the SO_ZEROCOPY
flag was introduced because send(2) ignores unknown message flags and
any legacy application which was accidentally passing the equivalent of
MSG_ZEROCOPY earlier should not see any new behaviour.
Before commit f214f915e7 ("tcp: enable MSG_ZEROCOPY"), a send(2) call
which passed the equivalent of MSG_ZEROCOPY without setting SO_ZEROCOPY
would succeed. However, after that commit, it fails with -ENOBUFS. So
it appears that the SO_ZEROCOPY flag fails to fulfill its intended
purpose. Fix it.
Fixes: f214f915e7 ("tcp: enable MSG_ZEROCOPY")
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking fixes from David Miller:
1) Must perform TXQ teardown before unregistering interfaces in
mac80211, from Toke Høiland-Jørgensen.
2) Don't allow creating mac80211_hwsim with less than one channel, from
Johannes Berg.
3) Division by zero in cfg80211, fix from Johannes Berg.
4) Fix endian issue in tipc, from Haiqing Bai.
5) BPF sockmap use-after-free fixes from Daniel Borkmann.
6) Spectre-v1 in mac80211_hwsim, from Jinbum Park.
7) Missing rhashtable_walk_exit() in tipc, from Cong Wang.
8) Revert kvzalloc() conversion of AF_PACKET, it breaks mmap() when
kvzalloc() tries to use kmalloc() pages. From Eric Dumazet.
9) Fix deadlock in hv_netvsc, from Dexuan Cui.
10) Do not restart timewait timer on RST, from Florian Westphal.
11) Fix double lwstate refcount grab in ipv6, from Alexey Kodanev.
12) Unsolicit report count handling is off-by-one, fix from Hangbin Liu.
13) Sleep-in-atomic in cadence driver, from Jia-Ju Bai.
14) Respect ttl-inherit in ip6 tunnel driver, from Hangbin Liu.
15) Use-after-free in act_ife, fix from Cong Wang.
16) Missing hold to meta module in act_ife, from Vlad Buslov.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (91 commits)
net: phy: sfp: Handle unimplemented hwmon limits and alarms
net: sched: action_ife: take reference to meta module
act_ife: fix a potential use-after-free
net/mlx5: Fix SQ offset in QPs with small RQ
tipc: correct spelling errors for tipc_topsrv_queue_evt() comments
tipc: correct spelling errors for struct tipc_bc_base's comment
bnxt_en: Do not adjust max_cp_rings by the ones used by RDMA.
bnxt_en: Clean up unused functions.
bnxt_en: Fix firmware signaled resource change logic in open.
sctp: not traverse asoc trans list if non-ipv6 trans exists for ipv6_flowlabel
sctp: fix invalid reference to the index variable of the iterator
net/ibm/emac: wrong emac_calc_base call was used by typo
net: sched: null actions array pointer before releasing action
vhost: fix VHOST_GET_BACKEND_FEATURES ioctl request definition
r8169: add support for NCube 8168 network card
ip6_tunnel: respect ttl inherit for ip6tnl
mac80211: shorten the IBSS debug messages
mac80211: don't Tx a deauth frame if the AP forbade Tx
mac80211: Fix station bandwidth setting after channel switch
mac80211: fix a race between restart and CSA flows
...
Helper bpf_msg_pull_data() mistakenly reuses variable 'offset' while
linearizing multiple scatterlist elements. Variable 'offset' is used
to find first starting scatterlist element
i.e. msg->data = sg_virt(&sg[first_sg]) + start - offset"
Use different variable name while linearizing multiple scatterlist
elements so that value contained in variable 'offset' won't get
overwritten.
Fixes: 015632bb30 ("bpf: sk_msg program helper bpf_sk_msg_pull_data")
Signed-off-by: Tushar Dave <tushar.n.dave@oracle.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
rtnl_unregister_all(PF_INET6) gets called from inet6_init in cases when
no handler has been registered for PF_INET6 yet, for example if
ip6_mr_init() fails. Abort and avoid a NULL pointer deref in that case.
Example of panic (triggered by faking a failure of
register_pernet_subsys):
general protection fault: 0000 [#1] PREEMPT SMP KASAN PTI
[...]
RIP: 0010:rtnl_unregister_all+0x17e/0x2a0
[...]
Call Trace:
? rtnetlink_net_init+0x250/0x250
? sock_unregister+0x103/0x160
? kernel_getsockopt+0x200/0x200
inet6_init+0x197/0x20d
Fixes: e2fddf5e96 ("[IPV6]: Make af_inet6 to check ip6_route_init return value.")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
When we perform the sg shift repair for the scatterlist ring, we
currently start out at i = first_sg + 1. However, this is not
correct since the first_sg could point to the sge sitting at slot
MAX_SKB_FRAGS - 1, and a subsequent i = MAX_SKB_FRAGS will access
the scatterlist ring (sg) out of bounds. Add the sk_msg_iter_var()
helper for iterating through the ring, and apply the same rule
for advancing to the next ring element as we do elsewhere. Later
work will use this helper also in other places.
Fixes: 015632bb30 ("bpf: sk_msg program helper bpf_sk_msg_pull_data")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
If first_sg and last_sg wraps around in the scatterlist ring, then we
need to account for that in the shift as well. E.g. crafting such msgs
where this is the case leads to a hang as shift becomes negative. E.g.
consider the following scenario:
first_sg := 14 |=> shift := -12 msg->sg_start := 10
last_sg := 3 | msg->sg_end := 5
round 1: i := 15, move_from := 3, sg[15] := sg[ 3]
round 2: i := 0, move_from := -12, sg[ 0] := sg[-12]
round 3: i := 1, move_from := -11, sg[ 1] := sg[-11]
round 4: i := 2, move_from := -10, sg[ 2] := sg[-10]
[...]
round 13: i := 11, move_from := -1, sg[ 2] := sg[ -1]
round 14: i := 12, move_from := 0, sg[ 2] := sg[ 0]
round 15: i := 13, move_from := 1, sg[ 2] := sg[ 1]
round 16: i := 14, move_from := 2, sg[ 2] := sg[ 2]
round 17: i := 15, move_from := 3, sg[ 2] := sg[ 3]
[...]
This means we will loop forever and never hit the msg->sg_end condition
to break out of the loop. When we see that the ring wraps around, then
the shift should be MAX_SKB_FRAGS - first_sg + last_sg - 1. Meaning,
the remainder slots from the tail of the ring and the head until last_sg
combined.
Fixes: 015632bb30 ("bpf: sk_msg program helper bpf_sk_msg_pull_data")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
In the current code, msg->data is set as sg_virt(&sg[i]) + start - offset
and msg->data_end relative to it as msg->data + bytes. Using iterator i
to point to the updated starting scatterlist element holds true for some
cases, however not for all where we'd end up pointing out of bounds. It
is /correct/ for these ones:
1) When first finding the starting scatterlist element (sge) where we
find that the page is already privately owned by the msg and where
the requested bytes and headroom fit into the sge's length.
However, it's /incorrect/ for the following ones:
2) After we made the requested area private and updated the newly allocated
page into first_sg slot of the scatterlist ring; when we find that no
shift repair of the ring is needed where we bail out updating msg->data
and msg->data_end. At that point i will point to last_sg, which in this
case is the next elem of first_sg in the ring. The sge at that point
might as well be invalid (e.g. i == msg->sg_end), which we use for
setting the range of sg_virt(&sg[i]). The correct one would have been
first_sg.
3) Similar as in 2) but when we find that a shift repair of the ring is
needed. In this case we fix up all sges and stop once we've reached the
end. In this case i will point to will point to the new msg->sg_end,
and the sge at that point will be invalid. Again here the requested
range sits in first_sg.
Fixes: 015632bb30 ("bpf: sk_msg program helper bpf_sk_msg_pull_data")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
While recently going over bpf_msg_pull_data(), I noticed three
issues which are fixed in here:
1) When we attempt to find the first scatterlist element (sge)
for the start offset, we add len to the offset before we check
for start < offset + len, whereas it should come after when
we iterate to the next sge to accumulate the offsets. For
example, given a start offset of 12 with a sge length of 8
for the first sge in the list would lead us to determine this
sge as the first sge thinking it covers first 16 bytes where
start is located, whereas start sits in subsequent sges so
we would end up pulling in the wrong data.
2) After figuring out the starting sge, we have a short-cut test
in !msg->sg_copy[i] && bytes <= len. This checks whether it's
not needed to make the page at the sge private where we can
just exit by updating msg->data and msg->data_end. However,
the length test is not fully correct. bytes <= len checks
whether the requested bytes (end - start offsets) fit into the
sge's length. The part that is missing is that start must not
be sge length aligned. Meaning, the start offset into the sge
needs to be accounted as well on top of the requested bytes
as otherwise we can access the sge out of bounds. For example
the sge could have length of 8, our requested bytes could have
length of 8, but at a start offset of 4, so we also would need
to pull in 4 bytes of the next sge, when we jump to the out
label we do set msg->data to sg_virt(&sg[i]) + start - offset
and msg->data_end to msg->data + bytes which would be oob.
3) The subsequent bytes < copy test for finding the last sge has
the same issue as in point 2) but also it tests for less than
rather than less or equal to. Meaning if the sge length is of
8 and requested bytes of 8 while having the start aligned with
the sge, we would unnecessarily go and pull in the next sge as
well to make it private.
Fixes: 015632bb30 ("bpf: sk_msg program helper bpf_sk_msg_pull_data")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Building the newly introduced BPF_PROG_TYPE_SK_REUSEPORT leads to
a compile time error when building with clang:
net/core/filter.o: In function `sk_reuseport_convert_ctx_access':
../net/core/filter.c:7284: undefined reference to `__compiletime_assert_7284'
It seems that clang has issues resolving hweight_long at compile
time. Since SK_FL_PROTO_MASK is a constant, we can use the interface
for known constant arguments which works fine with clang.
Fixes: 2dbb9b9e6d ("bpf: Introduce BPF_PROG_TYPE_SK_REUSEPORT")
Signed-off-by: Stefan Agner <stefan@agner.ch>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Pull IDA updates from Matthew Wilcox:
"A better IDA API:
id = ida_alloc(ida, GFP_xxx);
ida_free(ida, id);
rather than the cumbersome ida_simple_get(), ida_simple_remove().
The new IDA API is similar to ida_simple_get() but better named. The
internal restructuring of the IDA code removes the bitmap
preallocation nonsense.
I hope the net -200 lines of code is convincing"
* 'ida-4.19' of git://git.infradead.org/users/willy/linux-dax: (29 commits)
ida: Change ida_get_new_above to return the id
ida: Remove old API
test_ida: check_ida_destroy and check_ida_alloc
test_ida: Convert check_ida_conv to new API
test_ida: Move ida_check_max
test_ida: Move ida_check_leaf
idr-test: Convert ida_check_nomem to new API
ida: Start new test_ida module
target/iscsi: Allocate session IDs from an IDA
iscsi target: fix session creation failure handling
drm/vmwgfx: Convert to new IDA API
dmaengine: Convert to new IDA API
ppc: Convert vas ID allocation to new IDA API
media: Convert entity ID allocation to new IDA API
ppc: Convert mmu context allocation to new IDA API
Convert net_namespace to new IDA API
cb710: Convert to new IDA API
rsxx: Convert to new IDA API
osd: Convert to new IDA API
sd: Convert to new IDA API
...
Pull networking fixes from David Miller:
1) Fix races in IPVS, from Tan Hu.
2) Missing unbind in matchall classifier, from Hangbin Liu.
3) Missing act_ife action release, from Vlad Buslov.
4) Cure lockdep splats in ila, from Cong Wang.
5) veth queue leak on link delete, from Toshiaki Makita.
6) Disable isdn's IIOCDBGVAR ioctl, it exposes kernel addresses. From
Kees Cook.
7) RCU usage fixup in XDP, from Tariq Toukan.
8) Two TCP ULP fixes from Daniel Borkmann.
9) r8169 needs REALTEK_PHY as a Kconfig dependency, from Heiner
Kallweit.
10) Always take tcf_lock with BH disabled, otherwise we can deadlock
with rate estimator code paths. From Vlad Buslov.
11) Don't use MSI-X on RTL8106e r8169 chips, they don't resume properly.
From Jian-Hong Pan.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (41 commits)
ip6_vti: fix creating fallback tunnel device for vti6
ip_vti: fix a null pointer deferrence when create vti fallback tunnel
r8169: don't use MSI-X on RTL8106e
net: lan743x_ptp: convert to ktime_get_clocktai_ts64
net: sched: always disable bh when taking tcf_lock
ip6_vti: simplify stats handling in vti6_xmit
bpf: fix redirect to map under tail calls
r8169: add missing Kconfig dependency
tools/bpf: fix bpf selftest test_cgroup_storage failure
bpf, sockmap: fix sock_map_ctx_update_elem race with exist/noexist
bpf, sockmap: fix map elem deletion race with smap_stop_sock
bpf, sockmap: fix leakage of smap_psock_map_entry
tcp, ulp: fix leftover icsk_ulp_ops preventing sock from reattach
tcp, ulp: add alias for all ulp modules
bpf: fix a rcu usage warning in bpf_prog_array_copy_core()
samples/bpf: all XDP samples should unload xdp/bpf prog on SIGTERM
net/xdp: Fix suspicious RCU usage warning
net/mlx5e: Delete unneeded function argument
Documentation: networking: ti-cpsw: correct cbs parameters for Eth1 100Mb
isdn: Disable IIOCDBGVAR
...
Commits 109980b894 ("bpf: don't select potentially stale ri->map
from buggy xdp progs") and 7c30013133 ("bpf: fix ri->map_owner
pointer on bpf_prog_realloc") tried to mitigate that buggy programs
using bpf_redirect_map() helper call do not leave stale maps behind.
Idea was to add a map_owner cookie into the per CPU struct redirect_info
which was set to prog->aux by the prog making the helper call as a
proof that the map is not stale since the prog is implicitly holding
a reference to it. This owner cookie could later on get compared with
the program calling into BPF whether they match and therefore the
redirect could proceed with processing the map safely.
In (obvious) hindsight, this approach breaks down when tail calls are
involved since the original caller's prog->aux pointer does not have
to match the one from one of the progs out of the tail call chain,
and therefore the xdp buffer will be dropped instead of redirected.
A way around that would be to fix the issue differently (which also
allows to remove related work in fast path at the same time): once
the life-time of a redirect map has come to its end we use it's map
free callback where we need to wait on synchronize_rcu() for current
outstanding xdp buffers and remove such a map pointer from the
redirect info if found to be present. At that time no program is
using this map anymore so we simply invalidate the map pointers to
NULL iff they previously pointed to that instance while making sure
that the redirect path only reads out the map once.
Fixes: 97f91a7cf0 ("bpf: add bpf_redirect_map helper routine")
Fixes: 109980b894 ("bpf: don't select potentially stale ri->map from buggy xdp progs")
Reported-by: Sebastiano Miano <sebastiano.miano@polito.it>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
rdma.git merge resolution for the 4.19 merge window
Conflicts:
drivers/infiniband/core/rdma_core.c
- Use the rdma code and revise with the new spelling for
atomic_fetch_add_unless
drivers/nvme/host/rdma.c
- Replace max_sge with max_send_sge in new blk code
drivers/nvme/target/rdma.c
- Use the blk code and revise to use NULL for ib_post_recv when
appropriate
- Replace max_sge with max_recv_sge in new blk code
net/rds/ib_send.c
- Use the net code and revise to use NULL for ib_post_recv when
appropriate
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
-----BEGIN PGP SIGNATURE-----
iQFSBAABCAA8FiEEq68RxlopcLEwq+PEeb4+QwBBGIYFAltwm2geHHRvcnZhbGRz
QGxpbnV4LWZvdW5kYXRpb24ub3JnAAoJEHm+PkMAQRiGITkH/iSzkVhT2OxHoir0
mLVzTi7/Z17L0e/ELl7TvAC0iLFlWZKdlGR0g3b4/QpXLPmNK4HxiDRTQuWn8ke0
qDZyDq89HqLt+mpeFZ43PCd9oqV8CH2xxK3iCWReqv6bNnowGnRpSStlks4rDqWn
zURC/5sUh7TzEG4s997RrrpnyPeQWUlf/Mhtzg2/WvK2btoLWgu5qzjX1uFh3s7u
vaF2NXVJ3X03gPktyxZzwtO1SwLFS1jhwUXWBZ5AnoJ99ywkghQnkqS/2YpekNTm
wFk80/78sU+d91aAqO8kkhHj8VRrd+9SGnZ4mB2aZHwjZjGcics4RRtxukSfOQ+6
L47IdXo=
=sJkt
-----END PGP SIGNATURE-----
Merge tag 'v4.18' into rdma.git for-next
Resolve merge conflicts from the -rc cycle against the rdma.git tree:
Conflicts:
drivers/infiniband/core/uverbs_cmd.c
- New ifs added to ib_uverbs_ex_create_flow in -rc and for-next
- Merge removal of file->ucontext in for-next with new code in -rc
drivers/infiniband/core/uverbs_main.c
- for-next removed code from ib_uverbs_write() that was modified
in for-rc
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Addresses-Coverity-ID: 1472592 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
req->sdiag_family is a user-controlled value that's used as an array
index. Sanitize it after the bounds check to avoid speculative
out-of-bounds array access.
This also protects the sock_is_registered() call, so this removes the
sanitize call there.
Fixes: e978de7a6d ("net: socket: Fix potential spectre v1 gadget in sock_is_registered")
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: konrad.wilk@oracle.com
Cc: jamie.iles@oracle.com
Cc: liran.alon@oracle.com
Cc: stable@vger.kernel.org
Signed-off-by: Jeremy Cline <jcline@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann says:
====================
pull-request: bpf-next 2018-08-13
The following pull-request contains BPF updates for your *net-next* tree.
The main changes are:
1) Add driver XDP support for veth. This can be used in conjunction with
redirect of another XDP program e.g. sitting on NIC so the xdp_frame
can be forwarded to the peer veth directly without modification,
from Toshiaki.
2) Add a new BPF map type REUSEPORT_SOCKARRAY and prog type SK_REUSEPORT
in order to provide more control and visibility on where a SO_REUSEPORT
sk should be located, and the latter enables to directly select a sk
from the bpf map. This also enables map-in-map for application migration
use cases, from Martin.
3) Add a new BPF helper bpf_skb_ancestor_cgroup_id() that returns the id
of cgroup v2 that is the ancestor of the cgroup associated with the
skb at the ancestor_level, from Andrey.
4) Implement BPF fs map pretty-print support based on BTF data for regular
hash table and LRU map, from Yonghong.
5) Decouple the ability to attach BTF for a map from the key and value
pretty-printer in BPF fs, and enable further support of BTF for maps for
percpu and LPM trie, from Daniel.
6) Implement a better BPF sample of using XDP's CPU redirect feature for
load balancing SKB processing to remote CPU. The sample implements the
same XDP load balancing as Suricata does which is symmetric hash based
on IP and L4 protocol, from Jesper.
7) Revert adding NULL pointer check with WARN_ON_ONCE() in __xdp_return()'s
critical path as it is ensured that the allocator is present, from Björn.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
== Problem description ==
It's useful to be able to identify cgroup associated with skb in TC so
that a policy can be applied to this skb, and existing bpf_skb_cgroup_id
helper can help with this.
Though in real life cgroup hierarchy and hierarchy to apply a policy to
don't map 1:1.
It's often the case that there is a container and corresponding cgroup,
but there are many more sub-cgroups inside container, e.g. because it's
delegated to containerized application to control resources for its
subsystems, or to separate application inside container from infra that
belongs to containerization system (e.g. sshd).
At the same time it may be useful to apply a policy to container as a
whole.
If multiple containers like this are run on a host (what is often the
case) and many of them have sub-cgroups, it may not be possible to apply
per-container policy in TC with existing helpers such as
bpf_skb_under_cgroup or bpf_skb_cgroup_id:
* bpf_skb_cgroup_id will return id of immediate cgroup associated with
skb, i.e. if it's a sub-cgroup inside container, it can't be used to
identify container's cgroup;
* bpf_skb_under_cgroup can work only with one cgroup and doesn't scale,
i.e. if there are N containers on a host and a policy has to be
applied to M of them (0 <= M <= N), it'd require M calls to
bpf_skb_under_cgroup, and, if M changes, it'd require to rebuild &
load new BPF program.
== Solution ==
The patch introduces new helper bpf_skb_ancestor_cgroup_id that can be
used to get id of cgroup v2 that is an ancestor of cgroup associated
with skb at specified level of cgroup hierarchy.
That way admin can place all containers on one level of cgroup hierarchy
(what is a good practice in general and already used in many
configurations) and identify specific cgroup on this level no matter
what sub-cgroup skb is associated with.
E.g. if there is a cgroup hierarchy:
root/
root/container1/
root/container1/app11/
root/container1/app11/sub-app-a/
root/container1/app12/
root/container2/
root/container2/app21/
root/container2/app22/
root/container2/app22/sub-app-b/
, then having skb associated with root/container1/app11/sub-app-a/ it's
possible to get ancestor at level 1, what is container1 and apply policy
for this container, or apply another policy if it's container2.
Policies can be kept e.g. in a hash map where key is a container cgroup
id and value is an action.
Levels where container cgroups are created are usually known in advance
whether cgroup hierarchy inside container may be hard to predict
especially in case when its creation is delegated to containerized
application.
== Implementation details ==
The helper gets ancestor by walking parents up to specified level.
Another option would be to get different kind of "id" from
cgroup->ancestor_ids[level] and use it with idr_find() to get struct
cgroup for ancestor. But that would require radix lookup what doesn't
seem to be better (at least it's not obviously better).
Format of return value of the new helper is same as that of
bpf_skb_cgroup_id.
Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Extend gen_new_estimator() to also take stats_lock when re-assigning rate
estimator statistics pointer. (to be used by unlocked actions)
Rename 'stats_lock' to 'lock' and change argument description to explain
that it is now also used for control path.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch allows a BPF_PROG_TYPE_SK_REUSEPORT bpf prog to select a
SO_REUSEPORT sk from a BPF_MAP_TYPE_REUSEPORT_ARRAY introduced in
the earlier patch. "bpf_run_sk_reuseport()" will return -ECONNREFUSED
when the BPF_PROG_TYPE_SK_REUSEPORT prog returns SK_DROP.
The callers, in inet[6]_hashtable.c and ipv[46]/udp.c, are modified to
handle this case and return NULL immediately instead of continuing the
sk search from its hashtable.
It re-uses the existing SO_ATTACH_REUSEPORT_EBPF setsockopt to attach
BPF_PROG_TYPE_SK_REUSEPORT. The "sk_reuseport_attach_bpf()" will check
if the attaching bpf prog is in the new SK_REUSEPORT or the existing
SOCKET_FILTER type and then check different things accordingly.
One level of "__reuseport_attach_prog()" call is removed. The
"sk_unhashed() && ..." and "sk->sk_reuseport_cb" tests are pushed
back to "reuseport_attach_prog()" in sock_reuseport.c. sock_reuseport.c
seems to have more knowledge on those test requirements than filter.c.
In "reuseport_attach_prog()", after new_prog is attached to reuse->prog,
the old_prog (if any) is also directly freed instead of returning the
old_prog to the caller and asking the caller to free.
The sysctl_optmem_max check is moved back to the
"sk_reuseport_attach_filter()" and "sk_reuseport_attach_bpf()".
As of other bpf prog types, the new BPF_PROG_TYPE_SK_REUSEPORT is only
bounded by the usual "bpf_prog_charge_memlock()" during load time
instead of bounded by both bpf_prog_charge_memlock and sysctl_optmem_max.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This patch adds a BPF_PROG_TYPE_SK_REUSEPORT which can select
a SO_REUSEPORT sk from a BPF_MAP_TYPE_REUSEPORT_ARRAY. Like other
non SK_FILTER/CGROUP_SKB program, it requires CAP_SYS_ADMIN.
BPF_PROG_TYPE_SK_REUSEPORT introduces "struct sk_reuseport_kern"
to store the bpf context instead of using the skb->cb[48].
At the SO_REUSEPORT sk lookup time, it is in the middle of transiting
from a lower layer (ipv4/ipv6) to a upper layer (udp/tcp). At this
point, it is not always clear where the bpf context can be appended
in the skb->cb[48] to avoid saving-and-restoring cb[]. Even putting
aside the difference between ipv4-vs-ipv6 and udp-vs-tcp. It is not
clear if the lower layer is only ipv4 and ipv6 in the future and
will it not touch the cb[] again before transiting to the upper
layer.
For example, in udp_gro_receive(), it uses the 48 byte NAPI_GRO_CB
instead of IP[6]CB and it may still modify the cb[] after calling
the udp[46]_lib_lookup_skb(). Because of the above reason, if
sk->cb is used for the bpf ctx, saving-and-restoring is needed
and likely the whole 48 bytes cb[] has to be saved and restored.
Instead of saving, setting and restoring the cb[], this patch opts
to create a new "struct sk_reuseport_kern" and setting the needed
values in there.
The new BPF_PROG_TYPE_SK_REUSEPORT and "struct sk_reuseport_(kern|md)"
will serve all ipv4/ipv6 + udp/tcp combinations. There is no protocol
specific usage at this point and it is also inline with the current
sock_reuseport.c implementation (i.e. no protocol specific requirement).
In "struct sk_reuseport_md", this patch exposes data/data_end/len
with semantic similar to other existing usages. Together
with "bpf_skb_load_bytes()" and "bpf_skb_load_bytes_relative()",
the bpf prog can peek anywhere in the skb. The "bind_inany" tells
the bpf prog that the reuseport group is bind-ed to a local
INANY address which cannot be learned from skb.
The new "bind_inany" is added to "struct sock_reuseport" which will be
used when running the new "BPF_PROG_TYPE_SK_REUSEPORT" bpf prog in order
to avoid repeating the "bind INANY" test on
"sk_v6_rcv_saddr/sk->sk_rcv_saddr" every time a bpf prog is run. It can
only be properly initialized when a "sk->sk_reuseport" enabled sk is
adding to a hashtable (i.e. during "reuseport_alloc()" and
"reuseport_add_sock()").
The new "sk_select_reuseport()" is the main helper that the
bpf prog will use to select a SO_REUSEPORT sk. It is the only function
that can use the new BPF_MAP_TYPE_REUSEPORT_ARRAY. As mentioned in
the earlier patch, the validity of a selected sk is checked in
run time in "sk_select_reuseport()". Doing the check in
verification time is difficult and inflexible (consider the map-in-map
use case). The runtime check is to compare the selected sk's reuseport_id
with the reuseport_id that we want. This helper will return -EXXX if the
selected sk cannot serve the incoming request (e.g. reuseport_id
not match). The bpf prog can decide if it wants to do SK_DROP as its
discretion.
When the bpf prog returns SK_PASS, the kernel will check if a
valid sk has been selected (i.e. "reuse_kern->selected_sk != NULL").
If it does , it will use the selected sk. If not, the kernel
will select one from "reuse->socks[]" (as before this patch).
The SK_DROP and SK_PASS handling logic will be in the next patch.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This patch introduces a new map type BPF_MAP_TYPE_REUSEPORT_SOCKARRAY.
To unleash the full potential of a bpf prog, it is essential for the
userspace to be capable of directly setting up a bpf map which can then
be consumed by the bpf prog to make decision. In this case, decide which
SO_REUSEPORT sk to serve the incoming request.
By adding BPF_MAP_TYPE_REUSEPORT_SOCKARRAY, the userspace has total control
and visibility on where a SO_REUSEPORT sk should be located in a bpf map.
The later patch will introduce BPF_PROG_TYPE_SK_REUSEPORT such that
the bpf prog can directly select a sk from the bpf map. That will
raise the programmability of the bpf prog attached to a reuseport
group (a group of sk serving the same IP:PORT).
For example, in UDP, the bpf prog can peek into the payload (e.g.
through the "data" pointer introduced in the later patch) to learn
the application level's connection information and then decide which sk
to pick from a bpf map. The userspace can tightly couple the sk's location
in a bpf map with the application logic in generating the UDP payload's
connection information. This connection info contact/API stays within the
userspace.
Also, when used with map-in-map, the userspace can switch the
old-server-process's inner map to a new-server-process's inner map
in one call "bpf_map_update_elem(outer_map, &index, &new_reuseport_array)".
The bpf prog will then direct incoming requests to the new process instead
of the old process. The old process can finish draining the pending
requests (e.g. by "accept()") before closing the old-fds. [Note that
deleting a fd from a bpf map does not necessary mean the fd is closed]
During map_update_elem(),
Only SO_REUSEPORT sk (i.e. which has already been added
to a reuse->socks[]) can be used. That means a SO_REUSEPORT sk that is
"bind()" for UDP or "bind()+listen()" for TCP. These conditions are
ensured in "reuseport_array_update_check()".
A SO_REUSEPORT sk can only be added once to a map (i.e. the
same sk cannot be added twice even to the same map). SO_REUSEPORT
already allows another sk to be created for the same IP:PORT.
There is no need to re-create a similar usage in the BPF side.
When a SO_REUSEPORT is deleted from the "reuse->socks[]" (e.g. "close()"),
it will notify the bpf map to remove it from the map also. It is
done through "bpf_sk_reuseport_detach()" and it will only be called
if >=1 of the "reuse->sock[]" has ever been added to a bpf map.
The map_update()/map_delete() has to be in-sync with the
"reuse->socks[]". Hence, the same "reuseport_lock" used
by "reuse->socks[]" has to be used here also. Care has
been taken to ensure the lock is only acquired when the
adding sk passes some strict tests. and
freeing the map does not require the reuseport_lock.
The reuseport_array will also support lookup from the syscall
side. It will return a sock_gen_cookie(). The sock_gen_cookie()
is on-demand (i.e. a sk's cookie is not generated until the very
first map_lookup_elem()).
The lookup cookie is 64bits but it goes against the logical userspace
expectation on 32bits sizeof(fd) (and as other fd based bpf maps do also).
It may catch user in surprise if we enforce value_size=8 while
userspace still pass a 32bits fd during update. Supporting different
value_size between lookup and update seems unintuitive also.
We also need to consider what if other existing fd based maps want
to return 64bits value from syscall's lookup in the future.
Hence, reuseport_array supports both value_size 4 and 8, and
assuming user will usually use value_size=4. The syscall's lookup
will return ENOSPC on value_size=4. It will will only
return 64bits value from sock_gen_cookie() when user consciously
choose value_size=8 (as a signal that lookup is desired) which then
requires a 64bits value in both lookup and update.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
A later patch will introduce a BPF_MAP_TYPE_REUSEPORT_ARRAY which
allows a SO_REUSEPORT sk to be added to a bpf map. When a sk
is removed from reuse->socks[], it also needs to be removed from
the bpf map. Also, when adding a sk to a bpf map, the bpf
map needs to ensure it is indeed in a reuse->socks[].
Hence, reuseport_lock is needed by the bpf map to ensure its
map_update_elem() and map_delete_elem() operations are in-sync with
the reuse->socks[]. The BPF_MAP_TYPE_REUSEPORT_ARRAY map will only
acquire the reuseport_lock after ensuring the adding sk is already
in a reuseport group (i.e. reuse->socks[]). The map_lookup_elem()
will be lockless.
This patch also adds an ID to sock_reuseport. A later patch
will introduce BPF_PROG_TYPE_SK_REUSEPORT which allows
a bpf prog to select a sk from a bpf map. It is inflexible to
statically enforce a bpf map can only contain the sk belonging to
a particular reuse->socks[] (i.e. same IP:PORT) during the bpf
verification time. For example, think about the the map-in-map situation
where the inner map can be dynamically changed in runtime and the outer
map may have inner maps belonging to different reuseport groups.
Hence, when the bpf prog (in the new BPF_PROG_TYPE_SK_REUSEPORT
type) selects a sk, this selected sk has to be checked to ensure it
belongs to the requesting reuseport group (i.e. the group serving
that IP:PORT).
The "sk->sk_reuseport_cb" pointer cannot be used for this checking
purpose because the pointer value will change after reuseport_grow().
Instead of saving all checking conditions like the ones
preced calling "reuseport_add_sock()" and compare them everytime a
bpf_prog is run, a 32bits ID is introduced to survive the
reuseport_grow(). The ID is only acquired if any of the
reuse->socks[] is added to the newly introduced
"BPF_MAP_TYPE_REUSEPORT_ARRAY" map.
If "BPF_MAP_TYPE_REUSEPORT_ARRAY" is not used, the changes in this
patch is a no-op.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Although the actual cookie check "__cookie_v[46]_check()" does
not involve sk specific info, it checks whether the sk has recent
synq overflow event in "tcp_synq_no_recent_overflow()". The
tcp_sk(sk)->rx_opt.ts_recent_stamp is updated every second
when it has sent out a syncookie (through "tcp_synq_overflow()").
The above per sk "recent synq overflow event timestamp" works well
for non SO_REUSEPORT use case. However, it may cause random
connection request reject/discard when SO_REUSEPORT is used with
syncookie because it fails the "tcp_synq_no_recent_overflow()"
test.
When SO_REUSEPORT is used, it usually has multiple listening
socks serving TCP connection requests destinated to the same local IP:PORT.
There are cases that the TCP-ACK-COOKIE may not be received
by the same sk that sent out the syncookie. For example,
if reuse->socks[] began with {sk0, sk1},
1) sk1 sent out syncookies and tcp_sk(sk1)->rx_opt.ts_recent_stamp
was updated.
2) the reuse->socks[] became {sk1, sk2} later. e.g. sk0 was first closed
and then sk2 was added. Here, sk2 does not have ts_recent_stamp set.
There are other ordering that will trigger the similar situation
below but the idea is the same.
3) When the TCP-ACK-COOKIE comes back, sk2 was selected.
"tcp_synq_no_recent_overflow(sk2)" returns true. In this case,
all syncookies sent by sk1 will be handled (and rejected)
by sk2 while sk1 is still alive.
The userspace may create and remove listening SO_REUSEPORT sockets
as it sees fit. E.g. Adding new thread (and SO_REUSEPORT sock) to handle
incoming requests, old process stopping and new process starting...etc.
With or without SO_ATTACH_REUSEPORT_[CB]BPF,
the sockets leaving and joining a reuseport group makes picking
the same sk to check the syncookie very difficult (if not impossible).
The later patches will allow bpf prog more flexibility in deciding
where a sk should be located in a bpf map and selecting a particular
SO_REUSEPORT sock as it sees fit. e.g. Without closing any sock,
replace the whole bpf reuseport_array in one map_update() by using
map-in-map. Getting the syncookie check working smoothly across
socks in the same "reuse->socks[]" is important.
A partial solution is to set the newly added sk's ts_recent_stamp
to the max ts_recent_stamp of a reuseport group but that will require
to iterate through reuse->socks[] OR
pessimistically set it to "now - TCP_SYNCOOKIE_VALID" when a sk is
joining a reuseport group. However, neither of them will solve the
existing sk getting moved around the reuse->socks[] and that
sk may not have ts_recent_stamp updated, unlikely under continuous
synflood but not impossible.
This patch opts to treat the reuseport group as a whole when
considering the last synq overflow timestamp since
they are serving the same IP:PORT from the userspace
(and BPF program) perspective.
"synq_overflow_ts" is added to "struct sock_reuseport".
The tcp_synq_overflow() and tcp_synq_no_recent_overflow()
will update/check reuse->synq_overflow_ts if the sk is
in a reuseport group. Similar to the reuseport decision in
__inet_lookup_listener(), both sk->sk_reuseport and
sk->sk_reuseport_cb are tested for SO_REUSEPORT usage.
Update on "synq_overflow_ts" happens at roughly once
every second.
A synflood test was done with a 16 rx-queues and 16 reuseport sockets.
No meaningful performance change is observed. Before and
after the change is ~9Mpps in IPv4.
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
We need some mechanism to disable napi_direct on calling
xdp_return_frame_rx_napi() from some context.
When veth gets support of XDP_REDIRECT, it will redirects packets which
are redirected from other devices. On redirection veth will reuse
xdp_mem_info of the redirection source device to make return_frame work.
But in this case .ndo_xdp_xmit() called from veth redirection uses
xdp_mem_info which is not guarded by NAPI, because the .ndo_xdp_xmit()
is not called directly from the rxq which owns the xdp_mem_info.
This approach introduces a flag in bpf_redirect_info to indicate that
napi_direct should be disabled even when _rx_napi variant is used as
well as helper functions to use it.
A NAPI handler who wants to use this flag needs to call
xdp_set_return_frame_no_direct() before processing packets, and call
xdp_clear_return_frame_no_direct() after xdp_do_flush_map() before
exiting NAPI.
v4:
- Use bpf_redirect_info for storing the flag instead of xdp_mem_info to
avoid per-frame copy cost.
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
