When a queue(tfile) is detached through __tun_detach(), we move the
last enabled tfile to the position where detached one sit but don't
NULL out last position. We expect to synchronize the datapath through
tun->numqueues. Unfortunately, this won't work since we're lacking
sufficient mechanism to order or synchronize the access to
tun->numqueues.
To fix this, NULL out the last position during detaching and check
RCU protected tfile against NULL instead of checking tun->numqueues in
datapath.
Cc: YueHaibing <yuehaibing@huawei.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: weiyongjun (A) <weiyongjun1@huawei.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Fixes: c8d68e6be1 ("tuntap: multiqueue support")
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We need check if tun->numqueues is zero (e.g for the persist device)
before trying to use it for modular arithmetic.
Reported-by: Eric Dumazet <eric.dumazet@gmail.com>
Fixes: 96f84061620c6("tun: add eBPF based queue selection method")
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Update all users of eth_get_headlen to pass network device, fetch
network namespace from it and pass it down to the flow dissector.
This commit is a noop until administrator inserts BPF flow dissector
program.
Cc: Maxim Krasnyansky <maxk@qti.qualcomm.com>
Cc: Saeed Mahameed <saeedm@mellanox.com>
Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Cc: intel-wired-lan@lists.osuosl.org
Cc: Yisen Zhuang <yisen.zhuang@huawei.com>
Cc: Salil Mehta <salil.mehta@huawei.com>
Cc: Michael Chan <michael.chan@broadcom.com>
Cc: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
We prefer static_branch_unlikely() over static_key_false() these days.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit f2780d6d74 "tun: Add ioctl() SIOCGSKNS cmd to allow
obtaining net ns of tun device" it was missed that tun may change
its net ns, while net ns of socket remains the same as it was
created initially. SIOCGSKNS returns net ns of socket, so it is
not suitable for obtaining net ns of device.
We may have two tun devices with the same names in two net ns,
and in this case it's not possible to determ, which of them
fd refers to (TUNGETIFF will return the same name).
This patch adds new ioctl() cmd for obtaining net ns of a device.
Reported-by: Harald Albrecht <harald.albrecht@gmx.net>
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After the previous patch, all the callers of ndo_select_queue()
provide as a 'fallback' argument netdev_pick_tx.
The only exceptions are nested calls to ndo_select_queue(),
which pass down the 'fallback' available in the current scope
- still netdev_pick_tx.
We can drop such argument and replace fallback() invocation with
netdev_pick_tx(). This avoids an indirect call per xmit packet
in some scenarios (TCP syn, UDP unconnected, XDP generic, pktgen)
with device drivers implementing such ndo. It also clean the code
a bit.
Tested with ixgbe and CONFIG_FCOE=m
With pktgen using queue xmit:
threads vanilla patched
(kpps) (kpps)
1 2334 2428
2 4166 4278
4 7895 8100
v1 -> v2:
- rebased after helper's name change
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In my latest patch I missed one rcu_read_unlock(), in case
device is down.
Fixes: 4477138fa0 ("tun: properly test for IFF_UP")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Same reasons than the ones explained in commit 4179cb5a4c
("vxlan: test dev->flags & IFF_UP before calling netif_rx()")
netif_rx_ni() or napi_gro_frags() must be called under a strict contract.
At device dismantle phase, core networking clears IFF_UP
and flush_all_backlogs() is called after rcu grace period
to make sure no incoming packet might be in a cpu backlog
and still referencing the device.
A similar protocol is used for gro layer.
Most drivers call netif_rx() from their interrupt handler,
and since the interrupts are disabled at device dismantle,
netif_rx() does not have to check dev->flags & IFF_UP
Virtual drivers do not have this guarantee, and must
therefore make the check themselves.
Fixes: 1bd4978a88 ("tun: honor IFF_UP in tun_get_user()")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace set_current_state with __set_current_state since no memory
barrier is needed at this point.
Signed-off-by: Timur Celik <mail@timurcelik.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch moves setting of the current state into the loop. Otherwise
the task may end up in a busy wait loop if none of the break conditions
are met.
Signed-off-by: Timur Celik <mail@timurcelik.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
If the socket was created with socket(AF_PACKET, SOCK_RAW, 0),
skb->protocol will be unset, __skb_flow_dissect() will fail, and
skb_probe_transport_header() will fall back to the offset_hint, making
the resulting skb_transport_offset incorrect.
If, however, there is no transport header in the packet,
transport_header shouldn't be set to an arbitrary value.
Fix it by leaving the transport offset unset if it couldn't be found, to
be explicit rather than to fill it with some wrong value. It changes the
behavior, but if some code relied on the old behavior, it would be
broken anyway, as the old one is incorrect.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Call tun_set_real_num_queues() after the increment of tun->numqueues
since the former depends on it. Otherwise, the number of queues is not
correctly accounted for, which results to warnings similar to:
"vnet0 selects TX queue 11, but real number of TX queues is 11".
Fixes: 0b7959b625 ("tun: publish tfile after it's fully initialized")
Reported-and-tested-by: George Amanakis <gamanakis@gmail.com>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
BUG: unable to handle kernel NULL pointer dereference at 00000000000000d1
Call Trace:
? napi_gro_frags+0xa7/0x2c0
tun_get_user+0xb50/0xf20
tun_chr_write_iter+0x53/0x70
new_sync_write+0xff/0x160
vfs_write+0x191/0x1e0
__x64_sys_write+0x5e/0xd0
do_syscall_64+0x47/0xf0
entry_SYSCALL_64_after_hwframe+0x44/0xa9
I think there is a subtle race between sending a packet via tap and
attaching it:
CPU0: CPU1:
tun_chr_ioctl(TUNSETIFF)
tun_set_iff
tun_attach
rcu_assign_pointer(tfile->tun, tun);
tun_fops->write_iter()
tun_chr_write_iter
tun_napi_alloc_frags
napi_get_frags
napi->skb = napi_alloc_skb
tun_napi_init
netif_napi_add
napi->skb = NULL
napi->skb is NULL here
napi_gro_frags
napi_frags_skb
skb = napi->skb
skb_reset_mac_header(skb)
panic()
Move rcu_assign_pointer(tfile->tun) and rcu_assign_pointer(tun->tfiles) to
be the last thing we do in tun_attach(); this should guarantee that when we
call tun_get() we always get an initialized object.
v2 changes:
* remove extra napi_mutex locks/unlocks for napi operations
Reported-by: syzbot <syzkaller@googlegroups.com>
Fixes: 90e33d4594 ("tun: enable napi_gro_frags() for TUN/TAP driver")
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
tun_xdp_one() runs with local bh disabled. So there is no need to
disable preemption by calling get_cpu_ptr while updating stats. This
patch replaces the use of get_cpu_ptr() with this_cpu_ptr() as a
micro-optimization. Also removes related put_cpu_ptr call.
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Prashant Bhole <bhole_prashant_q7@lab.ntt.co.jp>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A follow-up patch will add a notifier type NETDEV_PRE_CHANGEADDR, which
allows vetoing of MAC address changes. One prominent path to that
notification is through dev_set_mac_address(). Therefore give this
function an extack argument, so that it can be packed together with the
notification. Thus a textual reason for rejection (or a warning) can be
communicated back to the user.
Signed-off-by: Petr Machata <petrm@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Several conflicts, seemingly all over the place.
I used Stephen Rothwell's sample resolutions for many of these, if not
just to double check my own work, so definitely the credit largely
goes to him.
The NFP conflict consisted of a bug fix (moving operations
past the rhashtable operation) while chaning the initial
argument in the function call in the moved code.
The net/dsa/master.c conflict had to do with a bug fix intermixing of
making dsa_master_set_mtu() static with the fixing of the tagging
attribute location.
cls_flower had a conflict because the dup reject fix from Or
overlapped with the addition of port range classifiction.
__set_phy_supported()'s conflict was relatively easy to resolve
because Andrew fixed it in both trees, so it was just a matter
of taking the net-next copy. Or at least I think it was :-)
Joe Stringer's fix to the handling of netns id 0 in bpf_sk_lookup()
intermixed with changes on how the sdif and caller_net are calculated
in these code paths in net-next.
The remaining BPF conflicts were largely about the addition of the
__bpf_md_ptr stuff in 'net' overlapping with adjustments and additions
to the relevant data structure where the MD pointer macros are used.
Signed-off-by: David S. Miller <davem@davemloft.net>
caller has guaranted that rxhash is not zero
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
tun flow entry 'updated' fields are written when receive
every packet. Thus if a flow is receiving packets from a
particular flow entry, it'll cause false-sharing with
all the other who has looked it up, so move it in its own
cache line
and update 'queue_index' and 'update' field only when
they are changed to reduce the cache false-sharing.
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Signed-off-by: Wang Li <wangli39@baidu.com>
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It's not supported right now (the goal of the initial patch was to support
'ip link del' only).
Before the patch:
$ ip link add foo type tun
[ 239.632660] BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
[snip]
[ 239.636410] RIP: 0010:register_netdevice+0x8e/0x3a0
This panic occurs because dev->netdev_ops is not set by tun_setup(). But to
have something usable, it will require more than just setting
netdev_ops.
Fixes: f019a7a594 ("tun: Implement ip link del tunXXX")
CC: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The userspace may need to control the carrier state.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: Didier Pallard <didier.pallard@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When writing packets to a descriptor associated with a combined queue, the
packets should end up on that queue.
Before this change all packets written to any descriptor associated with a
tap interface end up on rx-0, even when the descriptor is associated with a
different queue.
The rx traffic can be generated by either of the following.
1. a simple tap program which spins up multiple queues and writes packets
to each of the file descriptors
2. tx from a qemu vm with a tap multiqueue netdev
The queue for rx traffic can be observed by either of the following (done
on the hypervisor in the qemu case).
1. a simple netmap program which opens and reads from per-queue
descriptors
2. configuring RPS and doing per-cpu captures with rxtxcpu
Alternatively, if you printk() the return value of skb_get_rx_queue() just
before each instance of netif_receive_skb() in tun.c, you will get 65535
for every skb.
Calling skb_record_rx_queue() to set the rx queue to the queue_index fixes
the association between descriptor and rx queue.
Signed-off-by: Matthew Cover <matthew.cover@stackpath.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In order to cook skbs in the same way than Ethernet drivers,
it is probably better to not use GFP_KERNEL, but rather
use the GFP_ATOMIC and PFMEMALLOC mechanisms provided by
netdev_alloc_frag().
This would allow to use tun driver even in memory stress
situations, especially if swap is used over this tun channel.
Fixes: 90e33d4594 ("tun: enable napi_gro_frags() for TUN/TAP driver")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Petar Penkov <peterpenkov96@gmail.com>
Cc: Mahesh Bandewar <maheshb@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of constantly playing with the struct initializer
syntax trying to make gcc and CLang both happy, just clear
it out using memset().
>> drivers/net/tun.c:2503:42: warning: Using plain integer as NULL pointer
Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Thanks to the batched XDP buffs through msg_control. Instead of
calling put_page() for each page which involves a atomic operation,
let's batch them by record the last page that needs to be freed and
its refcnt count and free them in a batch.
Testpmd(virtio-user + vhost_net) + XDP_DROP shows 3.8% improvement.
Before: 4.71Mpps
After : 4.89Mpps
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The tun XDP sendmsg code path, unconditionally computes the symmetric
hash of each packet for RFS's sake, even when we could skip it. e.g.
when the device has a single queue.
This change adds the check already in-place for the skb sendmsg path
to avoid unneeded hashing.
The above gives small, but measurable, performance gain for VM xmit
path when zerocopy is not enabled.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Configuring generic network device parameters on tun will fail in
presence of IFLA_INFO_KIND attribute in IFLA_LINKINFO nested attribute
since tun_validate() always return failure.
This can be visualized with following ip-link(8) command sequences:
# ip link set dev tun0 group 100
# ip link set dev tun0 group 100 type tun
RTNETLINK answers: Invalid argument
with contrast to dummy and veth drivers:
# ip link set dev dummy0 group 100
# ip link set dev dummy0 type dummy
# ip link set dev veth0 group 100
# ip link set dev veth0 group 100 type veth
Fix by returning zero in tun_validate() when @data is NULL that is
always in case since rtnl_link_ops->maxtype is zero in tun driver.
Fixes: f019a7a594 ("tun: Implement ip link del tunXXX")
Signed-off-by: Serhey Popovych <serhe.popovych@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Because the function __skb_get_hash_symmetric always returns non-zero.
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Signed-off-by: Wang Li <wangli39@baidu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Minor conflict in net/core/rtnetlink.c, David Ahern's bug fix in 'net'
overlapped the renaming of a netlink attribute in net-next.
Signed-off-by: David S. Miller <davem@davemloft.net>
tun_napi_disable() and tun_napi_del() do not need
a pointer to the tun_struct
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Version bump conflict in batman-adv, take what's in net-next.
iavf conflict, adjustment of netdev_ops in net-next conflicting
with poll controller method removal in net.
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.
tun uses NAPI for TX completions, so we better let core
networking stack call the napi->poll() to avoid the capture.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch implement TUN_MSG_PTR msg_control type. This type allows
the caller to pass an array of XDP buffs to tuntap through ptr field
of the tun_msg_control. If an XDP program is attached, tuntap can run
XDP program directly. If not, tuntap will build skb and do a fast
receiving since part of the work has been done by vhost_net.
This will avoid lots of indirect calls thus improves the icache
utilization and allows to do XDP batched flushing when doing XDP
redirection.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch introduces to a new tun/tap specific msg_control:
#define TUN_MSG_UBUF 1
#define TUN_MSG_PTR 2
struct tun_msg_ctl {
int type;
void *ptr;
};
This allows us to pass different kinds of msg_control through
sendmsg(). The first supported type is ubuf (TUN_MSG_UBUF) which will
be used by the existed vhost_net zerocopy code. The second is XDP
buff, which allows vhost_net to pass XDP buff to TUN. This could be
used to implement accepting an array of XDP buffs from vhost_net in
the following patches.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch split out XDP logic into a single function. This make it to
be reused by XDP batching path in the following patch.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If we're sure not to go native XDP, there's no need for several things
like bh and rcu stuffs. So this patch introduces a helper to build skb
and hold page refcnt. When we found we will go through skb path, build
skb directly.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There's no need to duplicate page get logic in each action. So this
patch tries to get page and calculate the offset before processing XDP
actions (except for XDP_DROP), and undo them when meet errors (we
don't care the performance on errors). This will be used for factoring
out XDP logic.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch move the bh enabling a little bit earlier, this will be
used for factoring out the core XDP logic of tuntap.
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch introduces a new sock flag - SOCK_XDP. This will be used
for notifying the upper layer that XDP program is attached on the
lower socket, and requires for extra headroom.
TUN will be the first user.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull core signal handling updates from Eric Biederman:
"It was observed that a periodic timer in combination with a
sufficiently expensive fork could prevent fork from every completing.
This contains the changes to remove the need for that restart.
This set of changes is split into several parts:
- The first part makes PIDTYPE_TGID a proper pid type instead
something only for very special cases. The part starts using
PIDTYPE_TGID enough so that in __send_signal where signals are
actually delivered we know if the signal is being sent to a a group
of processes or just a single process.
- With that prep work out of the way the logic in fork is modified so
that fork logically makes signals received while it is running
appear to be received after the fork completes"
* 'siginfo-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace: (22 commits)
signal: Don't send signals to tasks that don't exist
signal: Don't restart fork when signals come in.
fork: Have new threads join on-going signal group stops
fork: Skip setting TIF_SIGPENDING in ptrace_init_task
signal: Add calculate_sigpending()
fork: Unconditionally exit if a fatal signal is pending
fork: Move and describe why the code examines PIDNS_ADDING
signal: Push pid type down into complete_signal.
signal: Push pid type down into __send_signal
signal: Push pid type down into send_signal
signal: Pass pid type into do_send_sig_info
signal: Pass pid type into send_sigio_to_task & send_sigurg_to_task
signal: Pass pid type into group_send_sig_info
signal: Pass pid and pid type into send_sigqueue
posix-timers: Noralize good_sigevent
signal: Use PIDTYPE_TGID to clearly store where file signals will be sent
pid: Implement PIDTYPE_TGID
pids: Move the pgrp and session pid pointers from task_struct to signal_struct
kvm: Don't open code task_pid in kvm_vcpu_ioctl
pids: Compute task_tgid using signal->leader_pid
...
0x3ff in tun_hashfn is mask of TUN_NUM_FLOW_ENTRIES, instead
of hardcode, define a macro to setup the relationship with
TUN_NUM_FLOW_ENTRIES
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When f_setown is called a pid and a pid type are stored. Replace the use
of PIDTYPE_PID with PIDTYPE_TGID as PIDTYPE_TGID goes to the entire thread
group. Replace the use of PIDTYPE_MAX with PIDTYPE_PID as PIDTYPE_PID now
is only for a thread.
Update the users of __f_setown to use PIDTYPE_TGID instead of
PIDTYPE_PID.
For now the code continues to capture task_pid (when task_tgid would
really be appropriate), and iterate on PIDTYPE_PID (even when type ==
PIDTYPE_TGID) out of an abundance of caution to preserve existing
behavior.
Oleg Nesterov suggested using the test to ensure we use PIDTYPE_PID
for tgid lookup also be used to avoid taking the tasklist lock.
Suggested-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
