Daniel Borkmann says:
====================
pull-request: bpf 2017-11-23
The following pull-request contains BPF updates for your *net* tree.
The main changes are:
1) Several BPF offloading fixes, from Jakub. Among others:
- Limit offload to cls_bpf and XDP program types only.
- Move device validation into the driver and don't make
any assumptions about the device in the classifier due
to shared blocks semantics.
- Don't pass offloaded XDP program into the driver when
it should be run in native XDP instead. Offloaded ones
are not JITed for the host in such cases.
- Don't destroy device offload state when moved to
another namespace.
- Revert dumping offload info into user space for now,
since ifindex alone is not sufficient. This will be
redone properly for bpf-next tree.
2) Fix test_verifier to avoid using bpf_probe_write_user()
helper in test cases, since it's dumping a warning into
kernel log which may confuse users when only running tests.
Switch to use bpf_trace_printk() instead, from Yonghong.
3) Several fixes for correcting ARG_CONST_SIZE_OR_ZERO semantics
before it becomes uabi, from Gianluca. More specifically:
- Add a type ARG_PTR_TO_MEM_OR_NULL that is used only
by bpf_csum_diff(), where the argument is either a
valid pointer or NULL. The subsequent ARG_CONST_SIZE_OR_ZERO
then enforces a valid pointer in case of non-0 size
or a valid pointer or NULL in case of size 0. Given
that, the semantics for ARG_PTR_TO_MEM in combination
with ARG_CONST_SIZE_OR_ZERO are now such that in case
of size 0, the pointer must always be valid and cannot
be NULL. This fix in semantics allows for bpf_probe_read()
to drop the recently added size == 0 check in the helper
that would become part of uabi otherwise once released.
At the same time we can then fix bpf_probe_read_str() and
bpf_perf_event_output() to use ARG_CONST_SIZE_OR_ZERO
instead of ARG_CONST_SIZE in order to fix recently
reported issues by Arnaldo et al, where LLVM optimizes
two boundary checks into a single one for unknown
variables where the verifier looses track of the variable
bounds and thus rejects valid programs otherwise.
4) A fix for the verifier for the case when it detects
comparison of two constants where the branch is guaranteed
to not be taken at runtime. Verifier will rightfully prune
the exploration of such paths, but we still pass the program
to JITs, where they would complain about using reserved
fields, etc. Track such dead instructions and sanitize
them with mov r0,r0. Rejection is not possible since LLVM
may generate them for valid C code and doesn't do as much
data flow analysis as verifier. For bpf-next we might
implement removal of such dead code and adjust branches
instead. Fix from Alexei.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Tuntap and similar devices can inject GSO packets. Accept type
VIRTIO_NET_HDR_GSO_UDP, even though not generating UFO natively.
Processes are expected to use feature negotiation such as TUNSETOFFLOAD
to detect supported offload types and refrain from injecting other
packets. This process breaks down with live migration: guest kernels
do not renegotiate flags, so destination hosts need to expose all
features that the source host does.
Partially revert the UFO removal from 182e0b6b5846~1..d9d30adf5677.
This patch introduces nearly(*) no new code to simplify verification.
It brings back verbatim tuntap UFO negotiation, VIRTIO_NET_HDR_GSO_UDP
insertion and software UFO segmentation.
It does not reinstate protocol stack support, hardware offload
(NETIF_F_UFO), SKB_GSO_UDP tunneling in SKB_GSO_SOFTWARE or reception
of VIRTIO_NET_HDR_GSO_UDP packets in tuntap.
To support SKB_GSO_UDP reappearing in the stack, also reinstate
logic in act_csum and openvswitch. Achieve equivalence with v4.13 HEAD
by squashing in commit 939912216f ("net: skb_needs_check() removes
CHECKSUM_UNNECESSARY check for tx.") and reverting commit 8d63bee643
("net: avoid skb_warn_bad_offload false positives on UFO").
(*) To avoid having to bring back skb_shinfo(skb)->ip6_frag_id,
ipv6_proxy_select_ident is changed to return a __be32 and this is
assigned directly to the frag_hdr. Also, SKB_GSO_UDP is inserted
at the end of the enum to minimize code churn.
Tested
Booted a v4.13 guest kernel with QEMU. On a host kernel before this
patch `ethtool -k eth0` shows UFO disabled. After the patch, it is
enabled, same as on a v4.13 host kernel.
A UFO packet sent from the guest appears on the tap device:
host:
nc -l -p -u 8000 &
tcpdump -n -i tap0
guest:
dd if=/dev/zero of=payload.txt bs=1 count=2000
nc -u 192.16.1.1 8000 < payload.txt
Direct tap to tap transmission of VIRTIO_NET_HDR_GSO_UDP succeeds,
packets arriving fragmented:
./with_tap_pair.sh ./tap_send_ufo tap0 tap1
(from https://github.com/wdebruij/kerneltools/tree/master/tests)
Changes
v1 -> v2
- simplified set_offload change (review comment)
- documented test procedure
Link: http://lkml.kernel.org/r/<CAF=yD-LuUeDuL9YWPJD9ykOZ0QCjNeznPDr6whqZ9NGMNF12Mw@mail.gmail.com>
Fixes: fb652fdfe8 ("macvlan/macvtap: Remove NETIF_F_UFO advertisement.")
Reported-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Florian reported a breakage with anycast routes due to commit
4832c30d54 ("net: ipv6: put host and anycast routes on device with
address"). Prior to this commit anycast routes were added against the
loopback device causing repetitive route entries with no insight into
why they existed. e.g.:
$ ip -6 ro ls table local type anycast
anycast 2001:db8:1:: dev lo proto kernel metric 0 pref medium
anycast 2001:db8:2:: dev lo proto kernel metric 0 pref medium
anycast fe80:: dev lo proto kernel metric 0 pref medium
anycast fe80:: dev lo proto kernel metric 0 pref medium
The point of commit 4832c30d54 is to add the routes using the device
with the address which is causing the route to be added. e.g.,:
$ ip -6 ro ls table local type anycast
anycast 2001:db8:1:: dev eth1 proto kernel metric 0 pref medium
anycast 2001:db8:2:: dev eth2 proto kernel metric 0 pref medium
anycast fe80:: dev eth2 proto kernel metric 0 pref medium
anycast fe80:: dev eth1 proto kernel metric 0 pref medium
For traffic to work as it did before, the dst device needs to be switched
to the loopback when the copy is created similar to local routes.
Fixes: 4832c30d54 ("net: ipv6: put host and anycast routes on device with address")
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With gcc-4.1.2:
net/smc/smc_core.c: In function ‘__smc_buf_create’:
net/smc/smc_core.c:567: warning: ‘bufsize’ may be used uninitialized in this function
Indeed, if the for-loop is never executed, bufsize is used
uninitialized. In addition, buf_desc is stored for later use, while it
is still a NULL pointer.
Before, error handling was done by checking if buf_desc is non-NULL.
The cleanup changed this to an error check, but forgot to update the
preinitialization of buf_desc to an error pointer.
Update the preinitializatin of buf_desc to fix this.
Fixes: b33982c3a6 ("net/smc: cleanup function __smc_buf_create()")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 3e034725c0 ("net/smc: common functions for RMBs and send buffers")
merged handling of SMC receive and send buffers. It introduced sk_buf_size
as merged start value for size determination. But since sk_buf_size is not
used at all, sk_sndbuf is erroneously used as start for rmb creation.
This patch makes sure, sk_buf_size is really used as intended, and
sk_rcvbuf is used as start value for rmb creation.
Fixes: 3e034725c0 ("net/smc: common functions for RMBs and send buffers")
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Hans Wippel <hwippel@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When the 'ignore_routes_with_linkdown' sysctl is set, we should not
consider linkdown nexthops during route lookup.
While the code correctly verifies that the initially selected route
('match') has a carrier, it does not perform the same check in the
subsequent multipath selection, resulting in a potential packet loss.
In case the chosen route does not have a carrier and the sysctl is set,
choose the initially selected route.
Fixes: 35103d1117 ("net: ipv6 sysctl option to ignore routes when nexthop link is down")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: David Ahern <dsahern@gmail.com>
Acked-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
If you flush (delete) a filter chain other than chain 0 (such as when
deleting the device), the kernel may run into a use-after-free. The
chain refcount must not be decremented unless we are sure we are done
with the chain.
To reproduce the bug, run:
ip link add dtest type dummy
tc qdisc add dev dtest ingress
tc filter add dev dtest chain 1 parent ffff: flower
ip link del dtest
Introduced in: commit f93e1cdcf4 ("net/sched: fix filter flushing"),
but unless you have KAsan or luck, you won't notice it until
commit 0dadc117ac ("cls_flower: use tcf_exts_get_net() before call_rcu()")
Fixes: f93e1cdcf4 ("net/sched: fix filter flushing")
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Roman Kapl <code@rkapl.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the current ARG_PTR_TO_MEM/ARG_PTR_TO_UNINIT_MEM semantics, an helper
argument can be NULL when the next argument type is ARG_CONST_SIZE_OR_ZERO
and the verifier can prove the value of this next argument is 0. However,
most helpers are just interested in handling <!NULL, 0>, so forcing them to
deal with <NULL, 0> makes the implementation of those helpers more
complicated for no apparent benefits, requiring them to explicitly handle
those corner cases with checks that bpf programs could start relying upon,
preventing the possibility of removing them later.
Solve this by making ARG_PTR_TO_MEM/ARG_PTR_TO_UNINIT_MEM never accept NULL
even when ARG_CONST_SIZE_OR_ZERO is set, and introduce a new argument type
ARG_PTR_TO_MEM_OR_NULL to explicitly deal with the NULL case.
Currently, the only helper that needs this is bpf_csum_diff_proto(), so
change arg1 and arg3 to this new type as well.
Also add a new battery of tests that explicitly test the
!ARG_PTR_TO_MEM_OR_NULL combination: all the current ones testing the
various <NULL, 0> variations are focused on bpf_csum_diff, so cover also
other helpers.
Signed-off-by: Gianluca Borello <g.borello@gmail.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Pull networking fixes from David Miller:
1) Fix a reference to a module parameter which was lost during the
GREv6 receive path rewrite, from Alexey Kodanev.
2) Fix deref before NULL check in ipheth, from Gustavo A. R. Silva.
3) RCU read lock imbalance in tun_build_skb(), from Xin Long.
4) Some stragglers from the mac80211 folks:
a) Timer conversions from Kees Cook
b) Fix some sequencing issue when cfg80211 is built statically,
from Johannes Berg
c) Memory leak in mac80211_hwsim, from Ben Hutchings.
5) Add new qmi_wwan device ID, from Sebastian Sjoholm.
6) Fix use after free in tipc, from Jon Maloy.
7) Missing kdoc in nfp driver, from Jakub Kicinski.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
nfp: flower: add missing kdoc
tipc: fix access of released memory
net: qmi_wwan: add Quectel BG96 2c7c:0296
mlxsw: spectrum: Do not try to create non-existing ports during unsplit
mac80211: properly free requested-but-not-started TX agg sessions
mac80211_hwsim: Fix memory leak in hwsim_new_radio_nl()
cfg80211: initialize regulatory keys/database later
mac80211: aggregation: Convert timers to use timer_setup()
nl80211: don't expose wdev->ssid for most interfaces
mac80211: Convert timers to use timer_setup()
net: vxge: Fix some indentation issues
net: ena: fix race condition between device reset and link up setup
r8169: use same RTL8111EVL green settings as in vendor driver
r8169: fix RTL8111EVL EEE and green settings
tun: fix rcu_read_lock imbalance in tun_build_skb
tcp: when scheduling TLP, time of RTO should account for current ACK
usbnet: ipheth: fix potential null pointer dereference in ipheth_carrier_set
gre6: use log_ecn_error module parameter in ip6_tnl_rcv()
state handling code cleanup from myself and some assorted CephFS fixes
from Jeff.
rbd now defaults to single-major=Y, lifting the limit of ~240 rbd
images per host for everyone.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
iQEcBAABCAAGBQJaEwyIAAoJEEp/3jgCEfOLjgYH/jKJbQ1yJFPyTVTTv/U9/xH2
kpHykEbzvvTT2TwNspbM9ZK4vSJPjYoHjL2qTRKxybuXYWYPxD2q6x+Z1iRP5G5N
4Py3RUZaagCSSgbUhfNl3VCbdki6cIKHHz1tHWBuO75kFEg03yZroozzc3SCKH8T
wHIa7UFxncDRroHMDiF5viF2tz4SfYSB0fd/Kev9qLJOiVr/lUTELfejlsu89ANT
6UvXPiTd9iifxQxjLV+2eQM4x5JImiDJUhMvcqfDlY2l85LzVCVTPXFnN4ZoEPlt
4NJj2SnnSQxSZLl1LwJC/gFYepdzW6qSxVqlpkAr0PvazZPushLpMA4AsKxWgVM=
=qsu2
-----END PGP SIGNATURE-----
Merge tag 'ceph-for-4.15-rc1' of git://github.com/ceph/ceph-client
Pull ceph updates from Ilya Dryomov:
"We have a set of file locking improvements from Zheng, rbd rw/ro state
handling code cleanup from myself and some assorted CephFS fixes from
Jeff.
rbd now defaults to single-major=Y, lifting the limit of ~240 rbd
images per host for everyone"
* tag 'ceph-for-4.15-rc1' of git://github.com/ceph/ceph-client:
rbd: default to single-major device number scheme
libceph: don't WARN() if user tries to add invalid key
rbd: set discard_alignment to zero
ceph: silence sparse endianness warning in encode_caps_cb
ceph: remove the bump of i_version
ceph: present consistent fsid, regardless of arch endianness
ceph: clean up spinlocking and list handling around cleanup_cap_releases()
rbd: get rid of rbd_mapping::read_only
rbd: fix and simplify rbd_ioctl_set_ro()
ceph: remove unused and redundant variable dropping
ceph: mark expected switch fall-throughs
ceph: -EINVAL on decoding failure in ceph_mdsc_handle_fsmap()
ceph: disable cached readdir after dropping positive dentry
ceph: fix bool initialization/comparison
ceph: handle 'session get evicted while there are file locks'
ceph: optimize flock encoding during reconnect
ceph: make lock_to_ceph_filelock() static
ceph: keep auth cap when inode has flocks or posix locks
* straggler timer conversions from Kees
* memory leak fix in hwsim
* fix some fallout from regdb changes if wireless is built-in
* also free aggregation sessions in startup state when station
goes away, to avoid crashing the timer
-----BEGIN PGP SIGNATURE-----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=jcSa
-----END PGP SIGNATURE-----
Merge tag 'mac80211-for-davem-2017-11-20' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211
Johannes Berg says:
====================
A few things:
* straggler timer conversions from Kees
* memory leak fix in hwsim
* fix some fallout from regdb changes if wireless is built-in
* also free aggregation sessions in startup state when station
goes away, to avoid crashing the timer
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
When the function tipc_group_filter_msg() finds that a member event
indicates that the member is leaving the group, it first deletes the
member instance, and then purges the message queue being handled
by the call. But the message queue is an aggregated field in the
just deleted item, leading the purge call to access freed memory.
We fix this by swapping the order of the two actions.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently device-bound programs are not able to run on the host
to save resources (host JIT is not invoked). Don't allow XDP
programs to be attached without the HW_MODE flag. In theory
if program is already translated for device offload the driver
should choose to offload it instead of loading it in the driver.
However, offloading translated program may still fail resulting
in device-bound program being run on the host.
Prevent this by refusing to attach device bound programs if
XDP_FLAGS_HW_MODE is not set.
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Quentin Monnet <quentin.monnet@netronome.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
With TC shared block changes we can't depend on correct netdev
pointer being available in cls_bpf. Move the device validation
to the driver. Core will only make sure that offloaded programs
are always attached in the driver (or in HW by the driver). We
trust that drivers which implement offload callbacks will perform
necessary checks.
Moving the checks to the driver is generally a useful thing,
in practice the check should be against a switchdev instance,
not a netdev, given that most ASICs will probably allow using
the same program on many ports.
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Quentin Monnet <quentin.monnet@netronome.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
When deleting a station or otherwise tearing down all aggregation
sessions, make sure to delete requested but not yet started ones,
to avoid the following scenario:
* session is requested, added to tid_start_tx[]
* ieee80211_ba_session_work() runs, gets past BLOCK_BA check
* ieee80211_sta_tear_down_BA_sessions() runs, locks &sta->ampdu_mlme.mtx,
e.g. while deleting the station - deleting all active sessions
* ieee80211_ba_session_work() continues since tear down flushes it, and
calls ieee80211_tx_ba_session_handle_start() for the new session, arms
the timer for it
* station deletion continues to __cleanup_single_sta() and frees the
session struct, while the timer is armed
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When cfg80211 is built as a module, everything is fine, and we
can keep the code as is; in fact, we have to, because there can
only be a single module_init().
When cfg80211 is built-in, however, it needs to initialize
before drivers (device_initcall/module_init), and thus used to
be at subsys_initcall(). I'd moved it to fs_initcall() earlier,
where it can remain. However, this is still too early because at
that point the key infrastructure hasn't been initialized yet,
so X.509 certificates can't be parsed yet.
To work around this problem, load the regdb keys only later in
a late_initcall(), at which point the necessary infrastructure
has been initialized.
Fixes: 90a53e4432 ("cfg80211: implement regdb signature checking")
Reported-by: Xiaolong Ye <xiaolong.ye@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
In preparation for unconditionally passing the struct timer_list pointer to
all timer callbacks, switch to using the new timer_setup() and from_timer()
to pass the timer pointer explicitly.
This removes the tid mapping array and expands the tid structures to
add a pointer back to the station, along with the tid index itself.
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: linux-wireless@vger.kernel.org
Cc: netdev@vger.kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
[switch tid variables to u8, the valid range is 0-15 at most,
initialize tid_tx->sta/tid properly]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
For mesh, this is simply wrong - there's no SSID, only the
mesh ID, so don't expose it at all.
For (P2P) client, it's wrong, because it exposes an internal
value that's only used when certain APIs are used.
For AP, it's actually the only correct case, so leave that.
All other interface types shouldn't be setting this anyway,
so there it won't change anything.
Fixes: b84e7a05f6 ("nl80211: send the NL80211_ATTR_SSID in nl80211_send_iface()")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
In preparation for unconditionally passing the struct timer_list pointer to
all timer callbacks, switch to using the new timer_setup() and from_timer()
to pass the timer pointer explicitly.
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: linux-wireless@vger.kernel.org
Cc: netdev@vger.kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Fix the TLP scheduling logic so that when scheduling a TLP probe, we
ensure that the estimated time at which an RTO would fire accounts for
the fact that ACKs indicating forward progress should push back RTO
times.
After the following fix:
df92c8394e ("tcp: fix xmit timer to only be reset if data ACKed/SACKed")
we had an unintentional behavior change in the following kind of
scenario: suppose the RTT variance has been very low recently. Then
suppose we send out a flight of N packets and our RTT is 100ms:
t=0: send a flight of N packets
t=100ms: receive an ACK for N-1 packets
The response before df92c8394e that was:
-> schedule a TLP for now + RTO_interval
The response after df92c8394e is:
-> schedule a TLP for t=0 + RTO_interval
Since RTO_interval = srtt + RTT_variance, this means that we have
scheduled a TLP timer at a point in the future that only accounts for
RTT_variance. If the RTT_variance term is small, this means that the
timer fires soon.
Before df92c8394e this would not happen, because in that code, when
we receive an ACK for a prefix of flight, we did:
1) Near the top of tcp_ack(), switch from TLP timer to RTO
at write_queue_head->paket_tx_time + RTO_interval:
if (icsk->icsk_pending == ICSK_TIME_LOSS_PROBE)
tcp_rearm_rto(sk);
2) In tcp_clean_rtx_queue(), update the RTO to now + RTO_interval:
if (flag & FLAG_ACKED) {
tcp_rearm_rto(sk);
3) In tcp_ack() after tcp_fastretrans_alert() switch from RTO
to TLP at now + RTO_interval:
if (icsk->icsk_pending == ICSK_TIME_RETRANS)
tcp_schedule_loss_probe(sk);
In df92c8394e we removed that 3-phase dance, and instead directly
set the TLP timer once: we set the TLP timer in cases like this to
write_queue_head->packet_tx_time + RTO_interval. So if the RTT
variance is small, then this means that this is setting the TLP timer
to fire quite soon. This means if the ACK for the tail of the flight
takes longer than an RTT to arrive (often due to delayed ACKs), then
the TLP timer fires too quickly.
Fixes: df92c8394e ("tcp: fix xmit timer to only be reset if data ACKed/SACKed")
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After commit 308edfdf15 ("gre6: Cleanup GREv6 receive path, call
common GRE functions") it's not used anywhere in the module, but
previously was used in ip6gre_rcv().
Fixes: 308edfdf15 ("gre6: Cleanup GREv6 receive path, call common GRE functions")
Signed-off-by: Alexey Kodanev <alexey.kodanev@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
- fix a number of races in the NFSv4+ state code.
- fix some shutdown crashes in multiple-network-namespace cases.
- relax our 4.1 session limits; if you've an artificially low limit
to the number of 4.1 clients that can mount simultaneously, try
upgrading.
-----BEGIN PGP SIGNATURE-----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=PBs+
-----END PGP SIGNATURE-----
Merge tag 'nfsd-4.15' of git://linux-nfs.org/~bfields/linux
Pull nfsd updates from Bruce Fields:
"Lots of good bugfixes, including:
- fix a number of races in the NFSv4+ state code
- fix some shutdown crashes in multiple-network-namespace cases
- relax our 4.1 session limits; if you've an artificially low limit
to the number of 4.1 clients that can mount simultaneously, try
upgrading"
* tag 'nfsd-4.15' of git://linux-nfs.org/~bfields/linux: (22 commits)
SUNRPC: Improve ordering of transport processing
nfsd: deal with revoked delegations appropriately
svcrdma: Enqueue after setting XPT_CLOSE in completion handlers
nfsd: use nfs->ns.inum as net ID
rpc: remove some BUG()s
svcrdma: Preserve CB send buffer across retransmits
nfds: avoid gettimeofday for nfssvc_boot time
fs, nfsd: convert nfs4_file.fi_ref from atomic_t to refcount_t
fs, nfsd: convert nfs4_cntl_odstate.co_odcount from atomic_t to refcount_t
fs, nfsd: convert nfs4_stid.sc_count from atomic_t to refcount_t
lockd: double unregister of inetaddr notifiers
nfsd4: catch some false session retries
nfsd4: fix cached replies to solo SEQUENCE compounds
sunrcp: make function _svc_create_xprt static
SUNRPC: Fix tracepoint storage issues with svc_recv and svc_rqst_status
nfsd: use ARRAY_SIZE
nfsd: give out fewer session slots as limit approaches
nfsd: increase DRC cache limit
nfsd: remove unnecessary nofilehandle checks
nfs_common: convert int to bool
...
Pull networking fixes from David Miller:
1) Revert regression inducing change to the IPSEC template resolver,
from Steffen Klassert.
2) Peeloffs can cause the wrong sk to be waken up in SCTP, fix from Xin
Long.
3) Min packet MTU size is wrong in cpsw driver, from Grygorii Strashko.
4) Fix build failure in netfilter ctnetlink, from Arnd Bergmann.
5) ISDN hisax driver checks pnp_irq() for errors incorrectly, from
Arvind Yadav.
6) Fix fealnx driver build failure on MIPS, from Huacai Chen.
7) Fix into leak in SCTP, the scope_id of socket addresses is not
always filled in. From Eric W. Biederman.
8) MTU inheritance between physical function and representor fix in nfp
driver, from Dirk van der Merwe.
9) Fix memory leak in rsi driver, from Colin Ian King.
10) Fix expiration and generation ID handling of cached ipv4 redirect
routes, from Xin Long.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (40 commits)
net: usb: hso.c: remove unneeded DRIVER_LICENSE #define
ibmvnic: fix dma_mapping_error call
ipvlan: NULL pointer dereference panic in ipvlan_port_destroy
route: also update fnhe_genid when updating a route cache
route: update fnhe_expires for redirect when the fnhe exists
sctp: set frag_point in sctp_setsockopt_maxseg correctly
rsi: fix memory leak on buf and usb_reg_buf
net/netlabel: Add list_next_rcu() in rcu_dereference().
nfp: remove false positive offloads in flower vxlan
nfp: register flower reprs for egress dev offload
nfp: inherit the max_mtu from the PF netdev
nfp: fix vlan receive MAC statistics typo
nfp: fix flower offload metadata flag usage
virto_net: remove empty file 'virtio_net.'
net/sctp: Always set scope_id in sctp_inet6_skb_msgname
fealnx: Fix building error on MIPS
isdn: hisax: Fix pnp_irq's error checking for setup_teles3
isdn: hisax: Fix pnp_irq's error checking for setup_sedlbauer_isapnp
isdn: hisax: Fix pnp_irq's error checking for setup_niccy
isdn: hisax: Fix pnp_irq's error checking for setup_ix1micro
...
Now when ip route flush cache and it turn out all fnhe_genid != genid.
If a redirect/pmtu icmp packet comes and the old fnhe is found and all
it's members but fnhe_genid will be updated.
Then next time when it looks up route and tries to rebind this fnhe to
the new dst, the fnhe will be flushed due to fnhe_genid != genid. It
causes this redirect/pmtu icmp packet acutally not to be applied.
This patch is to also reset fnhe_genid when updating a route cache.
Fixes: 5aad1de5ea ("ipv4: use separate genid for next hop exceptions")
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now when creating fnhe for redirect, it sets fnhe_expires for this
new route cache. But when updating the exist one, it doesn't do it.
It will cause this fnhe never to be expired.
Paolo already noticed it before, in Jianlin's test case, it became
even worse:
When ip route flush cache, the old fnhe is not to be removed, but
only clean it's members. When redirect comes again, this fnhe will
be found and updated, but never be expired due to fnhe_expires not
being set.
So fix it by simply updating fnhe_expires even it's for redirect.
Fixes: aee06da672 ("ipv4: use seqlock for nh_exceptions")
Reported-by: Jianlin Shi <jishi@redhat.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now in sctp_setsockopt_maxseg user_frag or frag_point can be set with
val >= 8 and val <= SCTP_MAX_CHUNK_LEN. But both checks are incorrect.
val >= 8 means frag_point can even be less than SCTP_DEFAULT_MINSEGMENT.
Then in sctp_datamsg_from_user(), when it's value is greater than cookie
echo len and trying to bundle with cookie echo chunk, the first_len will
overflow.
The worse case is when it's value is equal as cookie echo len, first_len
becomes 0, it will go into a dead loop for fragment later on. In Hangbin
syzkaller testing env, oom was even triggered due to consecutive memory
allocation in that loop.
Besides, SCTP_MAX_CHUNK_LEN is the max size of the whole chunk, it should
deduct the data header for frag_point or user_frag check.
This patch does a proper check with SCTP_DEFAULT_MINSEGMENT subtracting
the sctphdr and datahdr, SCTP_MAX_CHUNK_LEN subtracting datahdr when
setting frag_point via sockopt. It also improves sctp_setsockopt_maxseg
codes.
Suggested-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Reported-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add list_next_rcu() for fetching next list in rcu_deference safely.
Found with sparse in linux-next tree on tag next-20171116.
Signed-off-by: Tim Hansen <devtimhansen@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stable bugfixes:
- Revalidate "." and ".." correctly on open
- Avoid RCU usage in tracepoints
- Fix ugly referral attributes
- Fix a typo in nomigration mount option
- Revert "NFS: Move the flock open mode check into nfs_flock()"
Features:
- Implement a stronger send queue accounting system for NFS over RDMA
- Switch some atomics to the new refcount_t type
Other bugfixes and cleanups:
- Clean up access mode bits
- Remove special-case revalidations in nfs_opendir()
- Improve invalidating NFS over RDMA memory for async operations that time out
- Handle NFS over RDMA replies with a worqueue
- Handle NFS over RDMA sends with a workqueue
- Fix up replaying interrupted requests
- Remove dead NFS over RDMA definitions
- Update NFS over RDMA copyright information
- Be more consistent with bool initialization and comparisons
- Mark expected switch fall throughs
- Various sunrpc tracepoint cleanups
- Fix various OPEN races
- Fix a typo in nfs_rename()
- Use common error handling code in nfs_lock_and_join_request()
- Check that some structures are properly cleaned up during net_exit()
- Remove net pointer from dprintk()s
-----BEGIN PGP SIGNATURE-----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=2Zc1
-----END PGP SIGNATURE-----
Merge tag 'nfs-for-4.15-1' of git://git.linux-nfs.org/projects/anna/linux-nfs
Pull NFS client updates from Anna Schumaker:
"Stable bugfixes:
- Revalidate "." and ".." correctly on open
- Avoid RCU usage in tracepoints
- Fix ugly referral attributes
- Fix a typo in nomigration mount option
- Revert "NFS: Move the flock open mode check into nfs_flock()"
Features:
- Implement a stronger send queue accounting system for NFS over RDMA
- Switch some atomics to the new refcount_t type
Other bugfixes and cleanups:
- Clean up access mode bits
- Remove special-case revalidations in nfs_opendir()
- Improve invalidating NFS over RDMA memory for async operations that
time out
- Handle NFS over RDMA replies with a worqueue
- Handle NFS over RDMA sends with a workqueue
- Fix up replaying interrupted requests
- Remove dead NFS over RDMA definitions
- Update NFS over RDMA copyright information
- Be more consistent with bool initialization and comparisons
- Mark expected switch fall throughs
- Various sunrpc tracepoint cleanups
- Fix various OPEN races
- Fix a typo in nfs_rename()
- Use common error handling code in nfs_lock_and_join_request()
- Check that some structures are properly cleaned up during
net_exit()
- Remove net pointer from dprintk()s"
* tag 'nfs-for-4.15-1' of git://git.linux-nfs.org/projects/anna/linux-nfs: (62 commits)
NFS: Revert "NFS: Move the flock open mode check into nfs_flock()"
NFS: Fix typo in nomigration mount option
nfs: Fix ugly referral attributes
NFS: super: mark expected switch fall-throughs
sunrpc: remove net pointer from messages
nfs: remove net pointer from messages
sunrpc: exit_net cleanup check added
nfs client: exit_net cleanup check added
nfs/write: Use common error handling code in nfs_lock_and_join_requests()
NFSv4: Replace closed stateids with the "invalid special stateid"
NFSv4: nfs_set_open_stateid must not trigger state recovery for closed state
NFSv4: Check the open stateid when searching for expired state
NFSv4: Clean up nfs4_delegreturn_done
NFSv4: cleanup nfs4_close_done
NFSv4: Retry NFS4ERR_OLD_STATEID errors in layoutreturn
pNFS: Retry NFS4ERR_OLD_STATEID errors in layoutreturn-on-close
NFSv4: Don't try to CLOSE if the stateid 'other' field has changed
NFSv4: Retry CLOSE and DELEGRETURN on NFS4ERR_OLD_STATEID.
NFS: Fix a typo in nfs_rename()
NFSv4: Fix open create exclusive when the server reboots
...
Publishing of net pointer is not safe, use net->ns.inum as net ID
[ 171.391947] RPC: created new rpcb local clients
(rpcb_local_clnt: ..., rpcb_local_clnt4: ...) for net f00001e7
[ 171.767188] NFSD: starting 90-second grace period (net f00001e7)
Signed-off-by: Vasily Averin <vvs@virtuozzo.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Be sure that all_clients list initialized in net_init hook was return
to initial state.
Signed-off-by: Vasily Averin <vvs@virtuozzo.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Display information about the RPC procedure being requested in the
trace log. This sometimes critical information cannot always be
derived from other RPC trace entries.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
The rpc_task_begin trace point always display a task ID of zero.
Move the trace point call site so that it picks up the new task ID.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Signed-off-by: Gustavo A. R. Silva <garsilva@embeddedor.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Credit work contributed by Oracle engineers since 2014.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Clean up. This include should have been removed by
commit 23826c7aea ("xprtrdma: Serialize credit accounting again").
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Clean up: C-structure style XDR encoding and decoding logic has
been replaced over the past several merge windows on both the
client and server. These data structures are no longer used.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Lift the Send and LocalInv completion handlers out of soft IRQ mode
to make room for other work. Also, move the Send CQ to a different
CPU than the CPU where the Receive CQ is running, for improved
scalability.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Pull get_user_pages_fast() conversion from Al Viro:
"A bunch of places switched to get_user_pages_fast()"
* 'work.get_user_pages_fast' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
ceph: use get_user_pages_fast()
pvr2fs: use get_user_pages_fast()
atomisp: use get_user_pages_fast()
st: use get_user_pages_fast()
via_dmablit(): use get_user_pages_fast()
fsl_hypervisor: switch to get_user_pages_fast()
rapidio: switch to get_user_pages_fast()
vchiq_2835_arm: switch to get_user_pages_fast()
The sendctx circular queue now guarantees that xprtrdma cannot
overflow the Send Queue, so remove the remaining bits of the
original Send WQE counting mechanism.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
When an RPC Call includes a file data payload, that payload can come
from pages in the page cache, or a user buffer (for direct I/O).
If the payload can fit inline, xprtrdma includes it in the Send
using a scatter-gather technique. xprtrdma mustn't allow the RPC
consumer to re-use the memory where that payload resides before the
Send completes. Otherwise, the new contents of that memory would be
exposed by an HCA retransmit of the Send operation.
So, block RPC completion on Send completion, but only in the case
where a separate file data payload is part of the Send. This
prevents the reuse of that memory while it is still part of a Send
operation without an undue cost to other cases.
Waiting is avoided in the common case because typically the Send
will have completed long before the RPC Reply arrives.
These days, an RPC timeout will trigger a disconnect, which tears
down the QP. The disconnect flushes all waiting Sends. This bounds
the amount of time the reply handler has to wait for a Send
completion.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Invoke a common routine for releasing hardware resources (for
example, invalidating MRs). This needs to be done whether an
RPC Reply has arrived or the RPC was terminated early.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
We have one boolean flag in rpcrdma_req today. I'd like to add more
flags, so convert that boolean to a bit flag.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Problem statement:
Recently Sagi Grimberg <sagi@grimberg.me> observed that kernel RDMA-
enabled storage initiators don't handle delayed Send completion
correctly. If Send completion is delayed beyond the end of a ULP
transaction, the ULP may release resources that are still being used
by the HCA to complete a long-running Send operation.
This is a common design trait amongst our initiators. Most Send
operations are faster than the ULP transaction they are part of.
Waiting for a completion for these is typically unnecessary.
Infrequently, a network partition or some other problem crops up
where an ordering problem can occur. In NFS parlance, the RPC Reply
arrives and completes the RPC, but the HCA is still retrying the
Send WR that conveyed the RPC Call. In this case, the HCA can try
to use memory that has been invalidated or DMA unmapped, and the
connection is lost. If that memory has been re-used for something
else (possibly not related to NFS), and the Send retransmission
exposes that data on the wire.
Thus we cannot assume that it is safe to release Send-related
resources just because a ULP reply has arrived.
After some analysis, we have determined that the completion
housekeeping will not be difficult for xprtrdma:
- Inline Send buffers are registered via the local DMA key, and
are already left DMA mapped for the lifetime of a transport
connection, thus no additional handling is necessary for those
- Gathered Sends involving page cache pages _will_ need to
DMA unmap those pages after the Send completes. But like
inline send buffers, they are registered via the local DMA key,
and thus will not need to be invalidated
In addition, RPC completion will need to wait for Send completion
in the latter case. However, nearly always, the Send that conveys
the RPC Call will have completed long before the RPC Reply
arrives, and thus no additional latency will be accrued.
Design notes:
In this patch, the rpcrdma_sendctx object is introduced, and a
lock-free circular queue is added to manage a set of them per
transport.
The RPC client's send path already prevents sending more than one
RPC Call at the same time. This allows us to treat the consumer
side of the queue (rpcrdma_sendctx_get_locked) as if there is a
single consumer thread.
The producer side of the queue (rpcrdma_sendctx_put_locked) is
invoked only from the Send completion handler, which is a single
thread of execution (soft IRQ).
The only care that needs to be taken is with the tail index, which
is shared between the producer and consumer. Only the producer
updates the tail index. The consumer compares the head with the
tail to ensure that the a sendctx that is in use is never handed
out again (or, expressed more conventionally, the queue is empty).
When the sendctx queue empties completely, there are enough Sends
outstanding that posting more Send operations can result in a Send
Queue overflow. In this case, the ULP is told to wait and try again.
This introduces strong Send Queue accounting to xprtrdma.
As a final touch, Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
suggested a mechanism that does not require signaling every Send.
We signal once every N Sends, and perform SGE unmapping of N Send
operations during that one completion.
Reported-by: Sagi Grimberg <sagi@grimberg.me>
Suggested-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Commit 655fec6987 ("xprtrdma: Use gathered Send for large inline
messages") assumed that, since the zeroeth element of the Send SGE
array always pointed to req->rl_rdmabuf, it needed to be initialized
just once. This was a valid assumption because the Send SGE array
and rl_rdmabuf both live in the same rpcrdma_req.
In a subsequent patch, the Send SGE array will be separated from the
rpcrdma_req, so the zeroeth element of the SGE array needs to be
initialized every time.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Clean up: Make rpcrdma_prepare_send_sges() return a negative errno
instead of a bool. Soon callers will want distinct treatments of
different types of failures.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
When this function fails, it needs to undo the DMA mappings it's
done so far. Otherwise these are leaked.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Clean up. rpcrdma_prepare_hdr_sge() sets num_sge to one, then
rpcrdma_prepare_msg_sges() sets num_sge again to the count of SGEs
it added, plus one for the header SGE just mapped in
rpcrdma_prepare_hdr_sge(). This is confusing, and nails in an
assumption about when these functions are called.
Instead, maintain a running count that both functions can update
with just the number of SGEs they have added to the SGE array.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
We need to decode and save the incoming rdma_credits field _after_
we know that the direction of the message is "forward direction
Reply". Otherwise, the credits value in reverse direction Calls is
also used to update the forward direction credits.
It is safe to decode the rdma_credits field in rpcrdma_reply_handler
now that rpcrdma_reply_handler is single-threaded. Receives complete
in the same order as they were sent on the NFS server.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
I noticed that the soft IRQ thread looked pretty busy under heavy
I/O workloads. perf suggested one area that was expensive was the
queue_work() call in rpcrdma_wc_receive. That gave me some ideas.
Instead of scheduling a separate worker to process RPC Replies,
promote the Receive completion handler to IB_POLL_WORKQUEUE, and
invoke rpcrdma_reply_handler directly.
Note that the poll workqueue is single-threaded. In order to keep
memory invalidation from serializing all RPC Replies, handle any
necessary invalidation tasks in a separate multi-threaded workqueue.
This provides a two-tier scheme, similar to OS I/O interrupt
handlers: A fast interrupt handler that schedules the slow handler
and re-enables the interrupt, and a slower handler that is invoked
for any needed heavy lifting.
Benefits include:
- One less context switch for RPCs that don't register memory
- Receive completion handling is moved out of soft IRQ context to
make room for other users of soft IRQ
- The same CPU core now DMA syncs and XDR decodes the Receive buffer
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Clean up: I'd like to be able to invoke the tail of
rpcrdma_reply_handler in two different places. Split the tail out
into its own helper function.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
