Michael Chan [Mon, 9 Jul 2018 06:24:51 +0000 (02:24 -0400)]
bnxt_en: Do not modify max IRQ count after RDMA driver requests/frees IRQs.
Calling bnxt_set_max_func_irqs() to modify the max IRQ count requested or
freed by the RDMA driver is flawed. The max IRQ count is checked when
re-initializing the IRQ vectors and this can happen multiple times
during ifup or ethtool -L. If the max IRQ is reduced and the RDMA
driver is operational, we may not initailize IRQs correctly. This
problem shows up on VFs with very small number of MSIX.
There is no other logic that relies on the IRQ count excluding the ones
used by RDMA. So we fix it by just removing the call to subtract or
add the IRQs used by RDMA.
Fixes: a588e4580a7e ("bnxt_en: Add interface to support RDMA driver.") Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Mon, 9 Jul 2018 06:24:50 +0000 (02:24 -0400)]
bnxt_en: Support clearing of the IFF_BROADCAST flag.
Currently, the driver assumes IFF_BROADCAST is always set and always sets
the broadcast filter. Modify the code to set or clear the broadcast
filter according to the IFF_BROADCAST flag.
Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Mon, 9 Jul 2018 06:24:49 +0000 (02:24 -0400)]
bnxt_en: Always set output parameters in bnxt_get_max_rings().
The current code returns -ENOMEM and does not bother to set the output
parameters to 0 when no rings are available. Some callers, such as
bnxt_get_channels() will display garbage ring numbers when that happens.
Fix it by always setting the output parameters.
Fixes: 6e6c5a57fbe1 ("bnxt_en: Modify bnxt_get_max_rings() to support shared or non shared rings.") Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
If there aren't enough RX rings available, the driver will attempt to
use a single RX ring without the aggregation ring. If that also
fails, the BNXT_FLAG_AGG_RINGS flag is cleared but the other ring
parameters are not set consistently to reflect that. If more RX
rings become available at the next open, the RX rings will be in
an inconsistent state and may crash when freeing the RX rings.
Fix it by restoring the BNXT_FLAG_AGG_RINGS if not enough RX rings are
available to run without aggregation rings.
Fixes: bdbd1eb59c56 ("bnxt_en: Handle no aggregation ring gracefully.") Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
It is possible that OVS may set don’t care for DEI/CFI bit in
vlan_tci mask. Hence, checking for vlan_tci exact match will endup
in a vlan flow rejection.
This patch fixes the problem by checking for vlan_pcp and vid
separately, instead of checking for the entire vlan_tci.
Fixes: e85a9be93cf1 (bnxt_en: do not allow wildcard matches for L2 flows) Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
The following patchset contains Netfilter fixes for your net tree:
1) Missing module autoloadfor icmp and icmpv6 x_tables matches,
from Florian Westphal.
2) Possible non-linear access to TCP header from tproxy, from
Mate Eckl.
3) Do not allow rbtree to be used for single elements, this patch
moves all set backend into one single module since such thing
can only happen if hashtable module is explicitly blacklisted,
which should not ever be done.
4) Reject error and standard targets from nft_compat for sanity
reasons, they are never used from there.
5) Don't crash on double hashsize module parameter, from Andrey
Ryabinin.
6) Drop dst on skb before placing it in the fragmentation
reassembly queue, from Florian Westphal.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
netfilter: ipv6: nf_defrag: drop skb dst before queueing
Eric Dumazet reports:
Here is a reproducer of an annoying bug detected by syzkaller on our production kernel
[..]
./b78305423 enable_conntrack
Then :
sleep 60
dmesg | tail -10
[ 171.599093] unregister_netdevice: waiting for lo to become free. Usage count = 2
[ 181.631024] unregister_netdevice: waiting for lo to become free. Usage count = 2
[ 191.687076] unregister_netdevice: waiting for lo to become free. Usage count = 2
[ 201.703037] unregister_netdevice: waiting for lo to become free. Usage count = 2
[ 211.711072] unregister_netdevice: waiting for lo to become free. Usage count = 2
[ 221.959070] unregister_netdevice: waiting for lo to become free. Usage count = 2
Reproducer sends ipv6 fragment that hits nfct defrag via LOCAL_OUT hook.
skb gets queued until frag timer expiry -- 1 minute.
Normally nf_conntrack_reasm gets called during prerouting, so skb has
no dst yet which might explain why this wasn't spotted earlier.
Reported-by: Eric Dumazet <eric.dumazet@gmail.com> Reported-by: John Sperbeck <jsperbeck@google.com> Signed-off-by: Florian Westphal <fw@strlen.de> Tested-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
netfilter: nf_conntrack: Fix possible possible crash on module loading.
Loading the nf_conntrack module with doubled hashsize parameter, i.e.
modprobe nf_conntrack hashsize=12345 hashsize=12345
causes NULL-ptr deref.
If 'hashsize' specified twice, the nf_conntrack_set_hashsize() function
will be called also twice.
The first nf_conntrack_set_hashsize() call will set the
'nf_conntrack_htable_size' variable:
nf_conntrack_set_hashsize()
...
/* On boot, we can set this without any fancy locking. */
if (!nf_conntrack_htable_size)
return param_set_uint(val, kp);
But on the second invocation, the nf_conntrack_htable_size is already set,
so the nf_conntrack_set_hashsize() will take a different path and call
the nf_conntrack_hash_resize() function. Which will crash on the attempt
to dereference 'nf_conntrack_hash' pointer:
Fix this, by checking !nf_conntrack_hash instead of
!nf_conntrack_htable_size. nf_conntrack_hash will be initialized only
after the module loaded, so the second invocation of the
nf_conntrack_set_hashsize() won't crash, it will just reinitialize
nf_conntrack_htable_size again.
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
netfilter: nft_compat: explicitly reject ERROR and standard target
iptables-nft never requests these, but make this explicitly illegal.
If it were quested, kernel could oops as ->eval is NULL, furthermore,
the builtin targets have no owning module so its possible to rmmod
eb/ip/ip6_tables module even if they would be loaded.
Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Fixes: 05255b823a61 ("tcp: add TCP_ZEROCOPY_RECEIVE support for zerocopy receive") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Paolo Abeni [Fri, 6 Jul 2018 10:30:20 +0000 (12:30 +0200)]
ipfrag: really prevent allocation on netns exit
Setting the low threshold to 0 has no effect on frags allocation,
we need to clear high_thresh instead.
The code was pre-existent to commit 648700f76b03 ("inet: frags:
use rhashtables for reassembly units"), but before the above,
such assignment had a different role: prevent concurrent eviction
from the worker and the netns cleanup helper.
Fixes: 648700f76b03 ("inet: frags: use rhashtables for reassembly units") Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Lorenzo Colitti [Sat, 7 Jul 2018 07:31:40 +0000 (16:31 +0900)]
net: diag: Don't double-free TCP_NEW_SYN_RECV sockets in tcp_abort
When tcp_diag_destroy closes a TCP_NEW_SYN_RECV socket, it first
frees it by calling inet_csk_reqsk_queue_drop_and_and_put in
tcp_abort, and then frees it again by calling sock_gen_put.
Since tcp_abort only has one caller, and all the other codepaths
in tcp_abort don't free the socket, just remove the free in that
function.
Cc: David Ahern <dsa@cumulusnetworks.com>
Tested: passes Android sock_diag_test.py, which exercises this codepath Fixes: d7226c7a4dd1 ("net: diag: Fix refcnt leak in error path destroying socket") Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Reviewed-by: David Ahern <dsa@cumulusnetworks.com> Tested-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David Ahern [Sat, 7 Jul 2018 23:15:26 +0000 (16:15 -0700)]
net/ipv4: Set oif in fib_compute_spec_dst
Xin reported that icmp replies may not use the address on the device the
echo request is received if the destination address is broadcast. Instead
a route lookup is done without considering VRF context. Fix by setting
oif in flow struct to the master device if it is enslaved. That directs
the lookup to the VRF table. If the device is not enslaved, oif is still
0 so no affect.
Fixes: cd2fbe1b6b51 ("net: Use VRF device index for lookups on RX") Reported-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Otherwise we end up with attempting to send packets from down devices
or to send oversized packets, which may cause unexpected driver/device
behaviour. Generic XDP has already done this check, so reuse the logic
in native XDP.
====================
First three patches resolve issues found while testing sockhash and
reviewing code. Syzbot also found them about the same time as I was
working on fixes. The main issue is in the sockhash path we reduced
the scope of sk_callback lock but this meant we could get update and
close running in parallel so fix that here.
Then testing sk_msg and sk_skb programs together found that skb->dev
is not always assigned and some of the helpers were depending on this
to lookup max mtu. Fix this by using SKB_MAX_ALLOC when no MTU is
available.
Finally, Martin spotted that the sockmap code was still using the
qdisc skb cb structure. But I was sure we had fixed this long ago.
Looks like we missed it in a merge conflict resolution and then by
chance data_end offset was the same in both structures so everything
sort of continued to work even though it could break at any moment
if the structs ever change. So redo the conversion and this time
also convert the helpers.
v2: fix '0 files changed' issue in patches
====================
we added the routine bpf_compute_data_end_sk_skb() to compute the
correct data_end values, but this has since been lost. In kernel
v4.14 this was correct and the above patch was applied in it
entirety. Then when v4.14 was merged into v4.15-rc1 net-next tree
we lost the piece that renamed bpf_compute_data_pointers to the
new function bpf_compute_data_end_sk_skb. This was done here,
When it conflicted with the following rename patch,
6aaae2b6c433 ("bpf: rename bpf_compute_data_end into bpf_compute_data_pointers")
Finally, after a refactor I thought even the function
bpf_compute_data_end_sk_skb() was no longer needed and it was
erroneously removed.
However, we never reverted the sk_skb_convert_ctx_access() usage of
tcp_skb_cb which had been committed and survived the merge conflict.
Here we fix this by adding back the helper and *_data_end_sk_skb()
usage. Using the bpf_skc_data_end mapping is not correct because it
expects a qdisc_skb_cb object but at the sock layer this is not the
case. Even though it happens to work here because we don't overwrite
any data in-use at the socket layer and the cb structure is cleared
later this has potential to create some subtle issues. But, even
more concretely the filter.c access check uses tcp_skb_cb.
So when we use offset_of() to track down the byte offset we get 40 in
either case and everything continues to work. Fix this mess and use
correct structures its unclear how long this might actually work for
until someone moves the structs around.
Reported-by: Martin KaFai Lau <kafai@fb.com> Fixes: e1ea2f9856b7 ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net") Fixes: 6aaae2b6c433 ("bpf: rename bpf_compute_data_end into bpf_compute_data_pointers") Signed-off-by: John Fastabend <john.fastabend@gmail.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org>
John Fastabend [Thu, 5 Jul 2018 15:50:10 +0000 (08:50 -0700)]
bpf: sockmap, consume_skb in close path
Currently, when a sock is closed and the bpf_tcp_close() callback is
used we remove memory but do not free the skb. Call consume_skb() if
the skb is attached to the buffer.
John Fastabend [Thu, 5 Jul 2018 15:50:04 +0000 (08:50 -0700)]
bpf: sockhash, disallow bpf_tcp_close and update in parallel
After latest lock updates there is no longer anything preventing a
close and recvmsg call running in parallel. Additionally, we can
race update with close if we close a socket and simultaneously update
if via the BPF userspace API (note the cgroup ops are already run
with sock_lock held).
To resolve this take sock_lock in close and update paths.
Reported-by: syzbot+b680e42077a0d7c9a0c4@syzkaller.appspotmail.com Fixes: e9db4ef6bf4c ("bpf: sockhash fix omitted bucket lock in sock_close") Signed-off-by: John Fastabend <john.fastabend@gmail.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org>
John Fastabend [Thu, 5 Jul 2018 15:49:59 +0000 (08:49 -0700)]
bpf: fix sk_skb programs without skb->dev assigned
Multiple BPF helpers in use by sk_skb programs calculate the max
skb length using the __bpf_skb_max_len function. However, this
calculates the max length using the skb->dev pointer which can be
NULL when an sk_skb program is paired with an sk_msg program.
To force this a sk_msg program needs to redirect into the ingress
path of a sock with an attach sk_skb program. Then the the sk_skb
program would need to call one of the helpers that adjust the skb
size.
To fix the null ptr dereference use SKB_MAX_ALLOC size if no dev
is available.
====================
I missed fixing the error path in the sockhash code to align with
supporting socks in multiple maps. Simply checking if the psock is
present does not mean we can decrement the reference count because
it could be part of another map. Fix this by cleaning up the error
path so this situation does not happen.
====================
John Fastabend [Thu, 5 Jul 2018 15:05:56 +0000 (08:05 -0700)]
bpf: sockmap, error path can not release psock in multi-map case
The current code, in the error path of sock_hash_ctx_update_elem,
checks if the sock has a psock in the user data and if so decrements
the reference count of the psock. However, if the error happens early
in the error path we may have never incremented the psock reference
count and if the psock exists because the sock is in another map then
we may inadvertently decrement the reference count.
Fix this by making the error path only call smap_release_sock if the
error happens after the increment.
====================
net/sched: fix NULL dereference in 'goto chain' control action
in a couple of TC actions (i.e. csum and tunnel_key), the control action
is stored together with the action-specific configuration data.
This avoids a race condition (see [1]), but it causes a crash when 'goto
chain' is used with the above actions. Since this race condition is
tolerated on the other TC actions (it's present even on actions where the
spinlock is still used), storing the control action in the common area
should be acceptable for tunnel_key and csum as well.
net/sched: act_tunnel_key: fix NULL dereference when 'goto chain' is used
the control action in the common member of struct tcf_tunnel_key must be a
valid value, as it can contain the chain index when 'goto chain' is used.
Ensure that the control action can be read as x->tcfa_action, when x is a
pointer to struct tc_action and x->ops->type is TCA_ACT_TUNNEL_KEY, to
prevent the following command:
Fixes: d0f6dd8a914f ("net/sched: Introduce act_tunnel_key") Signed-off-by: Davide Caratti <dcaratti@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
net/sched: act_csum: fix NULL dereference when 'goto chain' is used
the control action in the common member of struct tcf_csum must be a valid
value, as it can contain the chain index when 'goto chain' is used. Ensure
that the control action can be read as x->tcfa_action, when x is a pointer
to struct tc_action and x->ops->type is TCA_ACT_CSUM, to prevent the
following command:
# tc filter add dev $h2 ingress protocol ip pref 1 handle 101 flower \
> $tcflags dst_mac $h2mac action csum ip or tcp or udp or sctp goto chain 1
from triggering a NULL pointer dereference when a matching packet is
received.
Fixes: 9c5f69bbd75a ("net/sched: act_csum: don't use spinlock in the fast path") Signed-off-by: Davide Caratti <dcaratti@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
net: macb: Allocate valid memory for TX and RX BD prefetch
GEM version in ZynqMP and most versions greater than r1p07 supports
TX and RX BD prefetch. The number of BDs that can be prefetched is a
HW configurable parameter. For ZynqMP, this parameter is 4.
When GEM DMA is accessing the last BD in the ring, even before the
BD is processed and the WRAP bit is noticed, it will have prefetched
BDs outside the BD ring. These will not be processed but it is
necessary to have accessible memory after the last BD. Especially
in cases where SMMU is used, memory locations immediately after the
last BD may not have translation tables triggering HRESP errors. Hence
always allocate extra BDs to accommodate for prefetch.
The value of tx/rx bd prefetch for any given SoC version is:
2 ^ (corresponding field in design config 10 register).
(value of this field >= 1)
Added a capability flag so that older IP versions that do not have
DCFG10 or this prefetch capability are not affected.
Signed-off-by: Harini Katakam <harini.katakam@xilinx.com> Reviewed-by: Claudiu Beznea <claudiu.beznea@microchip.com> Signed-off-by: David S. Miller <davem@davemloft.net>
rx ring is allocated for all queues in macb_alloc_consistent.
Free the same for all queues instead of just Q0.
Signed-off-by: Harini Katakam <harini.katakam@xilinx.com> Reviewed-by: Claudiu Beznea <claudiu.beznea@microchip.com> Signed-off-by: David S. Miller <davem@davemloft.net>
smc_release() calls a sock_put() for smc fallback sockets to cover
the passive closing sock_hold() in __smc_connect() and
smc_tcp_listen_work(). This does not make sense for sockets in state
SMC_LISTEN and SMC_INIT.
An SMC socket stays in state SMC_INIT if connect fails. The sock_put
in smc_connect_abort() does not cover all failures. Move it into
smc_connect_decline_fallback().
Fixes: ee9dfbef02d18 ("net/smc: handle sockopts forcing fallback") Reported-by: syzbot+3a0748c8f2f210c0ef9b@syzkaller.appspotmail.com Reported-by: syzbot+9e60d2428a42049a592a@syzkaller.appspotmail.com Signed-off-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
These two functions return the regular -EINVAL failure in the normal
code path, but return a nonstandard '-1' error otherwise, which gets
interpreted as -EPERM.
Let's change it to -EINVAL for the dummy functions as well.
Fixes: 4d4fd36126d6 ("net: bridge: Publish bridge accessor functions") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
cxgb4: assume flash part size to be 4MB, if it can't be determined
t4_get_flash_params() fails in a fatal fashion if the FLASH part isn't
one of the recognized parts. But this leads to desperate efforts to update
drivers when various FLASH parts which we are using suddenly become
unavailable and we need to substitute new FLASH parts. This has lead to
more than one Customer Field Emergency when a Customer has an old driver
and suddenly can't use newly shipped adapters.
This commit fixes this by simply assuming that the FLASH part is 4MB in
size if it can't be identified. Note that all Chelsio adapters will have
flash parts which are at least 4MB in size.
Signed-off-by: Casey Leedom <leedom@chelsio.com> Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Sat, 7 Jul 2018 10:49:02 +0000 (19:49 +0900)]
Merge branch 'tipc-dad-fixes'
Jon Maloy says:
====================
tipc: fixes in duplicate address discovery function
commit 25b0b9c4e835 ("tipc: handle collisions of 32-bit node address
hash values") introduced new functionality that has turned out to
contain several bugs and weaknesses.
We address those in this series.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Jon Maloy [Fri, 6 Jul 2018 18:10:06 +0000 (20:10 +0200)]
tipc: make function tipc_net_finalize() thread safe
The setting of the node address is not thread safe, meaning that
two discoverers may decide to set it simultanously, with a duplicate
entry in the name table as result. We fix that with this commit.
Fixes: 25b0b9c4e835 ("tipc: handle collisions of 32-bit node address hash values") Signed-off-by: Jon Maloy <jon.maloy@ericsson.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jon Maloy [Fri, 6 Jul 2018 18:10:05 +0000 (20:10 +0200)]
tipc: fix correct setting of message type in second discoverer
The duplicate address discovery protocol is not safe against two
discoverers running in parallel. The one executing first after the
trial period is over will set the node address and change its own
message type to DSC_REQ_MSG. The one executing last may find that the
node address is already set, and never change message type, with the
result that its links may never be established.
In this commmit we ensure that the message type always is set correctly
after the trial period is over.
Fixes: 25b0b9c4e835 ("tipc: handle collisions of 32-bit node address hash values") Signed-off-by: Jon Maloy <jon.maloy@ericsson.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jon Maloy [Fri, 6 Jul 2018 18:10:04 +0000 (20:10 +0200)]
tipc: correct discovery message handling during address trial period
With the duplicate address discovery protocol for tipc nodes addresses
we introduced a one second trial period before a node is allocated a
hash number to use as address.
Unfortunately, we miss to handle the case when a regular LINK REQUEST/
RESPONSE arrives from a cluster node during the trial period. Such
messages are not ignored as they should be, leading to links setup
attempts while the node still has no address.
Fixes: 25b0b9c4e835 ("tipc: handle collisions of 32-bit node address hash values") Signed-off-by: Jon Maloy <jon.maloy@ericsson.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jon Maloy [Fri, 6 Jul 2018 18:10:03 +0000 (20:10 +0200)]
tipc: fix wrong return value from function tipc_node_try_addr()
The function for checking if there is an node address conflict is
supposed to return a suggestion for a new address if it finds a
conflict, and zero otherwise. But in case the peer being checked
is previously unknown it does instead return a "suggestion" for
the checked address itself. This results in a DSC_TRIAL_FAIL_MSG
being sent unecessarily to the peer, and sometimes makes the trial
period starting over again.
Fixes: 25b0b9c4e835 ("tipc: handle collisions of 32-bit node address hash values") Signed-off-by: Jon Maloy <jon.maloy@ericsson.com> Signed-off-by: David S. Miller <davem@davemloft.net>
====================
ravb/sh_eth: fix sleep in atomic by reusing shared ethtool handlers
For ages trivial changes to RAVB and SuperH ethernet links by means of
standard 'ethtool' trigger a 'sleeping function called from invalid
context' bug, to visualize it on r8a7795 ULCB:
% ethtool -r eth0
BUG: sleeping function called from invalid context at kernel/locking/mutex.c:747
in_atomic(): 1, irqs_disabled(): 128, pid: 554, name: ethtool
INFO: lockdep is turned off.
irq event stamp: 0
hardirqs last enabled at (0): [<0000000000000000>] (null)
hardirqs last disabled at (0): [<ffff0000080e1d3c>] copy_process.isra.7.part.8+0x2cc/0x1918
softirqs last enabled at (0): [<ffff0000080e1d3c>] copy_process.isra.7.part.8+0x2cc/0x1918
softirqs last disabled at (0): [<0000000000000000>] (null)
CPU: 5 PID: 554 Comm: ethtool Not tainted 4.17.0-rc4-arm64-renesas+ #33
Hardware name: Renesas H3ULCB board based on r8a7795 ES2.0+ (DT)
Call trace:
dump_backtrace+0x0/0x198
show_stack+0x24/0x30
dump_stack+0xb8/0xf4
___might_sleep+0x1c8/0x1f8
__might_sleep+0x58/0x90
__mutex_lock+0x50/0x890
mutex_lock_nested+0x3c/0x50
phy_start_aneg_priv+0x38/0x180
phy_start_aneg+0x24/0x30
ravb_nway_reset+0x3c/0x68
dev_ethtool+0x3dc/0x2338
dev_ioctl+0x19c/0x490
sock_do_ioctl+0xe0/0x238
sock_ioctl+0x254/0x460
do_vfs_ioctl+0xb0/0x918
ksys_ioctl+0x50/0x80
sys_ioctl+0x34/0x48
__sys_trace_return+0x0/0x4
The root cause is that an attempt to modify ECMR and GECMR registers
only when RX/TX function is disabled was too overcomplicated in its
original implementation, also processing of an optional Link Change
interrupt added even more complexity, as a result the implementation
was error prone.
The new locking scheme is confirmed to be correct by dumping driver
specific and generic PHY framework function calls with aid of ftrace
while running more or less advanced tests.
Please note that sh_eth patches from the series were built-tested only.
On purpose I do not add Fixes tags, the reused PHY handlers were added
way later than the fixed problems were firstly found in the drivers.
Changes from v1 to v2:
* the original patches are split to bugfixes and enhancements only,
both v1 and v2 series are absolutely equal in total, thus I omit
description of changes in individual patches,
* the latter implies that there should be no strict need for retesting,
but because formally two series are different, I have to drop the tags
given by Geert and Andrew, please send your tags again.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
ravb: remove custom .set_link_ksettings from ethtool ops
The generic phy_ethtool_set_link_ksettings() function from phylib can
be used instead of in-house ravb_set_link_ksettings().
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
ravb: remove custom .get_link_ksettings from ethtool ops
The generic phy_ethtool_get_link_ksettings() function from phylib can be
used instead of in-house ravb_get_link_ksettings().
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
ravb: remove useless serialization in ravb_get_link_ksettings()
phy_ethtool_ksettings_get() call does not modify device state or device
driver state, hence there is no need to utilize a driver specific
spinlock.
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
The generic phy_ethtool_nway_reset() function from phylib can be used
instead of in-house ravb_nway_reset().
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
There is no need to call a heavyweight phy_start_aneg() for phy
auto-negotiation by ethtool, the phy is already initialized and
link auto-negotiation is started by calling phy_start() from
ravb_phy_start() when a network device is opened.
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
ravb: fix invalid context bug while changing link options by ethtool
The change fixes sleep in atomic context bug, which is encountered
every time when link settings are changed by ethtool.
Since commit 35b5f6b1a82b ("PHYLIB: Locking fixes for PHY I/O
potentially sleeping") phy_start_aneg() function utilizes a mutex
to serialize changes to phy state, however that helper function is
called in atomic context under a grabbed spinlock, because
phy_start_aneg() is called by phy_ethtool_ksettings_set() and by
replaced phy_ethtool_sset() helpers from phylib.
Now duplex mode setting is enforced in ravb_adjust_link() only, also
now RX/TX is disabled when link is put down or modifications to E-MAC
registers ECMR and GECMR are expected for both cases of checked and
ignored link status pin state from E-MAC interrupt handler.
Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper") Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
ravb: fix invalid context bug while calling auto-negotiation by ethtool
Since commit 35b5f6b1a82b ("PHYLIB: Locking fixes for PHY I/O
potentially sleeping") phy_start_aneg() function utilizes a mutex
to serialize changes to phy state, however the helper function is
called in atomic context.
The bug can be reproduced by running "ethtool -r" command, the bug
is reported if CONFIG_DEBUG_ATOMIC_SLEEP build option is enabled.
Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper") Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sh_eth: remove custom .set_link_ksettings from ethtool ops
The generic phy_ethtool_set_link_ksettings() function from phylib can
be used instead of in-house sh_eth_set_link_ksettings().
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sh_eth: remove custom .get_link_ksettings from ethtool ops
The generic phy_ethtool_get_link_ksettings() function from phylib can be
used instead of in-house sh_eth_get_link_ksettings().
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sh_eth: remove useless serialization in sh_eth_get_link_ksettings()
phy_ethtool_ksettings_get() call does not modify device state or device
driver state, hence there is no need to utilize a driver specific
spinlock.
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sh_eth: remove custom .nway_reset from ethtool ops
The generic phy_ethtool_nway_reset() function from phylib can be used
instead of in-house sh_eth_nway_reset().
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
There is no need to call a heavyweight phy_start_aneg() for phy
auto-negotiation by ethtool, the phy is already initialized and
link auto-negotiation is started by calling phy_start() from
sh_eth_phy_start() when a network device is opened.
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sh_eth: fix invalid context bug while changing link options by ethtool
The change fixes sleep in atomic context bug, which is encountered
every time when link settings are changed by ethtool.
Since commit 35b5f6b1a82b ("PHYLIB: Locking fixes for PHY I/O
potentially sleeping") phy_start_aneg() function utilizes a mutex
to serialize changes to phy state, however that helper function is
called in atomic context under a grabbed spinlock, because
phy_start_aneg() is called by phy_ethtool_ksettings_set() and by
replaced phy_ethtool_sset() helpers from phylib.
Now duplex mode setting is enforced in sh_eth_adjust_link() only,
also now RX/TX is disabled when link is put down or modifications
to E-MAC registers ECMR and GECMR are expected for both cases of
checked and ignored link status pin state from E-MAC interrupt handler.
For reference the change is a partial rework of commit 1e1b812bbe10
("sh_eth: fix handling of no LINK signal").
Fixes: dc19e4e5e02f ("sh: sh_eth: Add support ethtool") Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sh_eth: fix invalid context bug while calling auto-negotiation by ethtool
Since commit 35b5f6b1a82b ("PHYLIB: Locking fixes for PHY I/O
potentially sleeping") phy_start_aneg() function utilizes a mutex
to serialize changes to phy state, however the helper function is
called in atomic context.
The bug can be reproduced by running "ethtool -r" command, the bug
is reported if CONFIG_DEBUG_ATOMIC_SLEEP build option is enabled.
Fixes: dc19e4e5e02f ("sh: sh_eth: Add support ethtool") Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Signed-off-by: David S. Miller <davem@davemloft.net>
netfilter: nf_tables: place all set backends in one single module
This patch disallows rbtree with single elements, which is causing
problems with the recent timeout support. Before this patch, you
could opt out individual set representations per module, which is
just adding extra complexity.
netfilter: nf_tproxy: fix possible non-linear access to transport header
This patch fixes a silent out-of-bound read possibility that was present
because of the misuse of this function.
Mostly it was called with a struct udphdr *hp which had only the udphdr
part linearized by the skb_header_pointer, however
nf_tproxy_get_sock_v{4,6} uses it as a tcphdr pointer, so some reads for
tcp specific attributes may be invalid.
Fixes: a583636a83ea ("inet: refactor inet[6]_lookup functions to take skb") Signed-off-by: Máté Eckl <ecklm94@gmail.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
ipv4: Return EINVAL when ping_group_range sysctl doesn't map to user ns
The low and high values of the net.ipv4.ping_group_range sysctl were
being silently forced to the default disabled state when a write to the
sysctl contained GIDs that didn't map to the associated user namespace.
Confusingly, the sysctl's write operation would return success and then
a subsequent read of the sysctl would indicate that the low and high
values are the overflowgid.
This patch changes the behavior by clearly returning an error when the
sysctl write operation receives a GID range that doesn't map to the
associated user namespace. In such a situation, the previous value of
the sysctl is preserved and that range will be returned in a subsequent
read of the sysctl.
Signed-off-by: Tyler Hicks <tyhicks@canonical.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Igor Russkikh [Thu, 5 Jul 2018 14:01:09 +0000 (17:01 +0300)]
net: aquantia: vlan unicast address list correct handling
Setting up macvlan/macvtap networks over atlantic NIC results
in no traffic over these networks because ndo_set_rx_mode did
not listed UC MACs as registered in unicast filter.
Here we fix that taking into account maximum number of UC
filters supported by hardware. If more than MAX addresses were
registered, we just enable promisc and/or allmulti to pass
the traffic in.
We also remove MULTICAST_ADDRESS_MAX constant from aq_cfg since
thats not a configurable parameter at all.
Fixes: b21f502 ("net:ethernet:aquantia: Fix for multicast filter handling.") Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com> Signed-off-by: David S. Miller <davem@davemloft.net>
liquidio: make timeout HZ independent and readable
schedule_timeout_* takes a timeout in jiffies but the code currently is
passing in a constant which makes this timeout HZ dependent. So define
a constant with (hopefully) meaningful name and pass it through
msecs_to_jiffies() to fix the HZ dependency.
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
commit f21fb3ed364b ("Add support of Cavium Liquidio ethernet adapters") Acked-by: Derek Chickles <derek.chickles@caviumnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Thu, 5 Jul 2018 11:20:03 +0000 (20:20 +0900)]
Merge branch 'qrtr-Broadcasting-control-messages'
Arun Kumar Neelakantam says:
====================
net: qrtr: Broadcasting control messages
Allow messages only from control port to broadcast to avoid unnecessary
messages and reset the node to local router NODE ID in control messages
otherwise remote routers consider the packets as invalid and Drops it.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
net: qrtr: Reset the node and port ID of broadcast messages
All the control messages broadcast to remote routers are using
QRTR_NODE_BCAST instead of using local router NODE ID which cause
the packets to be dropped on remote router due to invalid NODE ID.
Signed-off-by: Arun Kumar Neelakantam <aneela@codeaurora.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Paul Moore [Wed, 4 Jul 2018 13:58:05 +0000 (09:58 -0400)]
ipv6: make ipv6_renew_options() interrupt/kernel safe
At present the ipv6_renew_options_kern() function ends up calling into
access_ok() which is problematic if done from inside an interrupt as
access_ok() calls WARN_ON_IN_IRQ() on some (all?) architectures
(x86-64 is affected). Example warning/backtrace is shown below:
The fix presented in this patch is to perform the userspace copy
earlier in the call chain such that it is only called when the option
data is actually coming from userspace; that place is
do_ipv6_setsockopt(). Not only does this solve the problem seen in
the backtrace above, it also allows us to simplify the code quite a
bit by removing ipv6_renew_options_kern() completely. We also take
this opportunity to cleanup ipv6_renew_options()/ipv6_renew_option()
a small amount as well.
This patch is heavily based on a rough patch by Al Viro. I've taken
his original patch, converted a kmemdup() call in do_ipv6_setsockopt()
to a memdup_user() call, made better use of the e_inval jump target in
the same function, and cleaned up the use ipv6_renew_option() by
ipv6_renew_options().
CC: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Paul Moore <paul@paul-moore.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Dan Carpenter [Wed, 4 Jul 2018 09:52:36 +0000 (12:52 +0300)]
qed: off by one in qed_parse_mcp_trace_buf()
If format_idx == s_mcp_trace_meta.formats_num then we read one element
beyond the end of the s_mcp_trace_meta.formats[] array.
Fixes: 50bc60cb155c ("qed*: Utilize FW 8.33.11.0") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Acked-by: Tomer Tayar <Tomer.Tayar@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Taeung Song <treeze.taeung@gmail.com> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Taeung Song [Wed, 4 Jul 2018 13:36:38 +0000 (22:36 +0900)]
samples/bpf: Check the error of write() and read()
test_task_rename() and test_urandom_read()
can be failed during write() and read(),
So check the result of them.
Reviewed-by: David Laight <David.Laight@ACULAB.COM> Signed-off-by: Taeung Song <treeze.taeung@gmail.com> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Taeung Song [Wed, 4 Jul 2018 13:36:37 +0000 (22:36 +0900)]
samples/bpf: Check the result of system()
To avoid the below build warning message,
use new generate_load() checking the return value.
ignoring return value of ‘system’, declared with attribute warn_unused_result
And it also refactors the duplicate code of both
test_perf_event_all_cpu() and test_perf_event_task()
Cc: Teng Qin <qinteng@fb.com> Signed-off-by: Taeung Song <treeze.taeung@gmail.com> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Taeung Song [Wed, 4 Jul 2018 13:36:36 +0000 (22:36 +0900)]
samples/bpf: add missing <linux/if_vlan.h>
This fixes build error regarding redefinition:
CLANG-bpf samples/bpf/parse_varlen.o
samples/bpf/parse_varlen.c:111:8: error: redefinition of 'vlan_hdr'
struct vlan_hdr {
^
./include/linux/if_vlan.h:38:8: note: previous definition is here
So remove duplicate 'struct vlan_hdr' in sample code and include if_vlan.h
Signed-off-by: Taeung Song <treeze.taeung@gmail.com> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Configuration information read at driver load can become stale after it is
updated. Mark information as not valid and re-populate when this happens.
Signed-off-by: Denis Bolotin <denis.bolotin@cavium.com> Signed-off-by: Ariel Elior <ariel.elior@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Currently also the pause flags are removed from phydev->supported because
they're not included in PHY_DEFAULT_FEATURES. I don't think this is
intended, especially when considering that this function can be called
via phy_set_max_speed() anywhere in a driver. Change the masking to mask
out only the values we're going to change. In addition remove the
misleading comment, job of this small function is just to adjust the
supported and advertised speeds.
Fixes: f3a6bd393c2c ("phylib: Add phy_set_max_speed helper") Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
smsc75xx: Add workaround for gigabit link up hardware errata.
In certain conditions, the device may not be able to link in gigabit mode. This software workaround ensures that the device will not enter the failure state.
Fixes: d0cad871703b898a442e4049c532ec39168e5b57 ("SMSC75XX USB 2.0 Gigabit Ethernet Devices") Signed-off-by: Yuiko Oshino <yuiko.oshino@microchip.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Xin Long [Tue, 3 Jul 2018 08:30:47 +0000 (16:30 +0800)]
sctp: fix the issue that pathmtu may be set lower than MINSEGMENT
After commit b6c5734db070 ("sctp: fix the handling of ICMP Frag Needed
for too small MTUs"), sctp_transport_update_pmtu would refetch pathmtu
from the dst and set it to transport's pathmtu without any check.
The new pathmtu may be lower than MINSEGMENT if the dst is obsolete and
updated by .get_dst() in sctp_transport_update_pmtu. In this case, it
could have a smaller MTU as well, and thus we should validate it
against MINSEGMENT instead.
Syzbot reported a warning in sctp_mtu_payload caused by this.
This patch refetches the pathmtu by calling sctp_dst_mtu where it does
the check against MINSEGMENT.
v1->v2:
- refetch the pathmtu by calling sctp_dst_mtu instead as Marcelo's
suggestion.
Fixes: b6c5734db070 ("sctp: fix the handling of ICMP Frag Needed for too small MTUs") Reported-by: syzbot+f0d9d7cba052f9344b03@syzkaller.appspotmail.com Suggested-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Signed-off-by: Xin Long <lucien.xin@gmail.com> Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David Ahern [Tue, 3 Jul 2018 21:36:21 +0000 (14:36 -0700)]
net/ipv6: Revert attempt to simplify route replace and append
NetworkManager likes to manage linklocal prefix routes and does so with
the NLM_F_APPEND flag, breaking attempts to simplify the IPv6 route
code and by extension enable multipath routes with device only nexthops.
Revert f34436a43092 and these followup patches: 6eba08c3626b ("ipv6: Only emit append events for appended routes"). ce45bded6435 ("mlxsw: spectrum_router: Align with new route replace logic") 53b562df8c20 ("mlxsw: spectrum_router: Allow appending to dev-only routes")
Update the fib_tests cases to reflect the old behavior.
Fixes: f34436a43092 ("net/ipv6: Simplify route replace and appending into multipath route") Signed-off-by: David Ahern <dsahern@gmail.com>
gen_stats: Fix netlink stats dumping in the presence of padding
The gen_stats facility will add a header for the toplevel nlattr of type
TCA_STATS2 that contains all stats added by qdisc callbacks. A reference
to this header is stored in the gnet_dump struct, and when all the
per-qdisc callbacks have finished adding their stats, the length of the
containing header will be adjusted to the right value.
However, on architectures that need padding (i.e., that don't set
CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS), the padding nlattr is added
before the stats, which means that the stored pointer will point to the
padding, and so when the header is fixed up, the result is just a very
big padding nlattr. Because most qdiscs also supply the legacy TCA_STATS
struct, this problem has been mostly invisible, but we exposed it with
the netlink attribute-based statistics in CAKE.
Fix the issue by fixing up the stored pointer if it points to a padding
nlattr.
Tested-by: Pete Heist <pete@heistp.net> Tested-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk> Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Tue, 3 Jul 2018 14:29:29 +0000 (23:29 +0900)]
Merge tag 'wireless-drivers-for-davem-2018-07-03' of git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/wireless-drivers
Kalle Valo says:
====================
wireless-drivers fixes for 4.18
First set of fixes for 4.18 and for numerous drivers. Something to mention
about is the wcn36xx fix which makes it possible to compile with gcc older than
4.4 (though I'm not sure if we even support those anymore).
qtnfmac
* coverity fix for a new commit in v4.18-rc1
rtlwifi
* fix kernel oops during driver removal
* fix firmware image corruption for rtl8821ae
brcmfmac
* fix crash if there's no firmware image
mwifiex
* a revert and a better fix for a new commit v4.18-rc1
mt7601u
* fix a recent regression about unnecessary warning about avg_rssi
wcn36xx
* convert testmode.c to plain ASCII
ath10k
* fix a firmware crash during bandwidth change
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
The current code does not inspect the return value of skb_to_sgvec. This
can cause a nullptr kernel panic when the malformed sgvec is passed into
the crypto request.
Checking the return value of skb_to_sgvec and skipping decryption if it
is negative fixes this problem.
Fixes: c46234ebb4d1 ("tls: RX path for ktls") Acked-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Doron Roberts-Kedes <doronrk@fb.com> Signed-off-by: David S. Miller <davem@davemloft.net>
lib: rhashtable: Correct self-assignment in rhashtable.c
In file lib/rhashtable.c line 777, skip variable is assigned to
itself. The following error was observed:
lib/rhashtable.c:777:41: warning: explicitly assigning value of
variable of type 'int' to itself [-Wself-assign] error, forbidden
warning: rhashtable.c:777
This error was found when compiling with Clang 6.0. Change it to iter->skip.
Signed-off-by: Rishabh Bhatnagar <rishabhb@codeaurora.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Magnus Karlsson [Fri, 29 Jun 2018 07:48:20 +0000 (09:48 +0200)]
xsk: fix potential race in SKB TX completion code
There is a potential race in the TX completion code for the SKB
case. One process enters the sendmsg code of an AF_XDP socket in order
to send a frame. The execution eventually trickles down to the driver
that is told to send the packet. However, it decides to drop the
packet due to some error condition (e.g., rings full) and frees the
SKB. This will trigger the SKB destructor and a completion will be
sent to the AF_XDP user space through its
single-producer/single-consumer queues.
At the same time a TX interrupt has fired on another core and it
dispatches the TX completion code in the driver. It does its HW
specific things and ends up freeing the SKB associated with the
transmitted packet. This will trigger the SKB destructor and a
completion will be sent to the AF_XDP user space through its
single-producer/single-consumer queues. With a pseudo call stack, it
would look like this:
Core 1:
sendmsg() being called in the application
netdev_start_xmit()
Driver entered through ndo_start_xmit
Driver decides to free the SKB for some reason (e.g., rings full)
Destructor of SKB called
xskq_produce_addr() is called to signal completion to user space
Core 2:
TX completion irq
NAPI loop
Driver irq handler for TX completions
Frees the SKB
Destructor of SKB called
xskq_produce_addr() is called to signal completion to user space
We now have a violation of the single-producer/single-consumer
principle for our queues as there are two threads trying to produce at
the same time on the same queue.
Fixed by introducing a spin_lock in the destructor. In regards to the
performance, I get around 1.74 Mpps for txonly before and after the
introduction of the spinlock. There is of course some impact due to
the spin lock but it is in the less significant digits that are too
noisy for me to measure. But let us say that the version without the
spin lock got 1.745 Mpps in the best case and the version with 1.735
Mpps in the worst case, then that would mean a maximum drop in
performance of 0.5%.
Fixes: 35fcde7f8deb ("xsk: support for Tx") Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Magnus Karlsson [Fri, 29 Jun 2018 07:48:19 +0000 (09:48 +0200)]
samples/bpf: deal with EBUSY return code from sendmsg in xdpsock sample
Sendmsg in the SKB path of AF_XDP can now return EBUSY when a packet
was discarded and completed by the driver. Just ignore this message
in the sample application.
Fixes: b4b8faa1ded7 ("samples/bpf: sample application and documentation for AF_XDP sockets") Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com> Reported-by: Pavel Odintsov <pavel@fastnetmon.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Magnus Karlsson [Fri, 29 Jun 2018 07:48:18 +0000 (09:48 +0200)]
xsk: frame could be completed more than once in SKB path
Fixed a bug in which a frame could be completed more than once
when an error was returned from dev_direct_xmit(). The code
erroneously retried sending the message leading to multiple
calls to the SKB destructor and therefore multiple completions
of the same buffer to user space.
The error code in this case has been changed from EAGAIN to EBUSY
in order to tell user space that the sending of the packet failed
and the buffer has been return to user space through the completion
queue.
Fixes: 35fcde7f8deb ("xsk: support for Tx") Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com> Reported-by: Pavel Odintsov <pavel@fastnetmon.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Magnus Karlsson [Fri, 29 Jun 2018 07:48:17 +0000 (09:48 +0200)]
xsk: fix potential lost completion message in SKB path
The code in xskq_produce_addr erroneously checked if there
was up to LAZY_UPDATE_THRESHOLD amount of space in the completion
queue. It only needs to check if there is one slot left in the
queue. This bug could under some circumstances lead to a WARN_ON_ONCE
being triggered and the completion message to user space being lost.
Fixes: 35fcde7f8deb ("xsk: support for Tx") Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com> Reported-by: Pavel Odintsov <pavel@fastnetmon.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md
Pull MD fixes from Shaohua Li:
"Two small fixes for MD:
- an error handling fix from me
- a recover bug fix for raid10 from BingJing"
* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md:
md/raid10: fix that replacement cannot complete recovery after reassemble
MD: cleanup resources in failure
1) Verify netlink attributes properly in nf_queue, from Eric Dumazet.
2) Need to bump memory lock rlimit for test_sockmap bpf test, from
Yonghong Song.
3) Fix VLAN handling in lan78xx driver, from Dave Stevenson.
4) Fix uninitialized read in nf_log, from Jann Horn.
5) Fix raw command length parsing in mlx5, from Alex Vesker.
6) Cleanup loopback RDS connections upon netns deletion, from Sowmini
Varadhan.
7) Fix regressions in FIB rule matching during create, from Jason A.
Donenfeld and Roopa Prabhu.
8) Fix mpls ether type detection in nfp, from Pieter Jansen van Vuuren.
9) More bpfilter build fixes/adjustments from Masahiro Yamada.
10) Fix XDP_{TX,REDIRECT} flushing in various drivers, from Jesper
Dangaard Brouer.
11) fib_tests.sh file permissions were broken, from Shuah Khan.
12) Make sure BH/preemption is disabled in data path of mac80211, from
Denis Kenzior.
13) Don't ignore nla_parse_nested() return values in nl80211, from
Johannes berg.
14) Properly account sock objects ot kmemcg, from Shakeel Butt.
15) Adjustments to setting bpf program permissions to read-only, from
Daniel Borkmann.
16) TCP Fast Open key endianness was broken, it always took on the host
endiannness. Whoops. Explicitly make it little endian. From Yuching
Cheng.
17) Fix prefix route setting for link local addresses in ipv6, from
David Ahern.
18) Potential Spectre v1 in zatm driver, from Gustavo A. R. Silva.
19) Various bpf sockmap fixes, from John Fastabend.
20) Use after free for GRO with ESP, from Sabrina Dubroca.
21) Passing bogus flags to crypto_alloc_shash() in ipv6 SR code, from
Eric Biggers.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (87 commits)
qede: Adverstise software timestamp caps when PHC is not available.
qed: Fix use of incorrect size in memcpy call.
qed: Fix setting of incorrect eswitch mode.
qed: Limit msix vectors in kdump kernel to the minimum required count.
ipvlan: call dev_change_flags when ipvlan mode is reset
ipv6: sr: fix passing wrong flags to crypto_alloc_shash()
net: fix use-after-free in GRO with ESP
tcp: prevent bogus FRTO undos with non-SACK flows
bpf: sockhash, add release routine
bpf: sockhash fix omitted bucket lock in sock_close
bpf: sockmap, fix smap_list_map_remove when psock is in many maps
bpf: sockmap, fix crash when ipv6 sock is added
net: fib_rules: bring back rule_exists to match rule during add
hv_netvsc: split sub-channel setup into async and sync
net: use dev_change_tx_queue_len() for SIOCSIFTXQLEN
atm: zatm: Fix potential Spectre v1
s390/qeth: consistently re-enable device features
s390/qeth: don't clobber buffer on async TX completion
s390/qeth: avoid using is_multicast_ether_addr_64bits on (u8 *)[6]
s390/qeth: fix race when setting MAC address
...
qede: Adverstise software timestamp caps when PHC is not available.
When ptp clock is not available for a PF (e.g., higher PFs in NPAR mode),
get-tsinfo() callback should return the software timestamp capabilities
instead of returning the error.
Fixes: 4c55215c ("qede: Add driver support for PTP") Signed-off-by: Sudarsana Reddy Kalluru <Sudarsana.Kalluru@cavium.com> Signed-off-by: Michal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Use the correct size value while copying chassis/port id values.
Fixes: 6ad8c632e ("qed: Add support for query/config dcbx.") Signed-off-by: Sudarsana Reddy Kalluru <Sudarsana.Kalluru@cavium.com> Signed-off-by: Michal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
By default, driver sets the eswitch mode incorrectly as VEB (virtual
Ethernet bridging).
Need to set VEB eswitch mode only when sriov is enabled, and it should be
to set NONE by default. The patch incorporates this change.
Fixes: 0fefbfbaa ("qed*: Management firmware - notifications and defaults") Signed-off-by: Sudarsana Reddy Kalluru <Sudarsana.Kalluru@cavium.com> Signed-off-by: Michal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
qed: Limit msix vectors in kdump kernel to the minimum required count.
Memory size is limited in the kdump kernel environment. Allocation of more
msix-vectors (or queues) consumes few tens of MBs of memory, which might
lead to the kdump kernel failure.
This patch adds changes to limit the number of MSI-X vectors in kdump
kernel to minimum required value (i.e., 2 per engine).
Fixes: fe56b9e6a ("qed: Add module with basic common support") Signed-off-by: Sudarsana Reddy Kalluru <Sudarsana.Kalluru@cavium.com> Signed-off-by: Michal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Hangbin Liu [Sun, 1 Jul 2018 08:21:21 +0000 (16:21 +0800)]
ipvlan: call dev_change_flags when ipvlan mode is reset
After we change the ipvlan mode from l3 to l2, or vice versa, we only
reset IFF_NOARP flag, but don't flush the ARP table cache, which will
cause eth->h_dest to be equal to eth->h_source in ipvlan_xmit_mode_l2().
Then the message will not come out of host.
Here is the reproducer on local host:
ip link set eth1 up
ip addr add 192.168.1.1/24 dev eth1
ip link add link eth1 ipvlan1 type ipvlan mode l3
ip netns add net1
ip link set ipvlan1 netns net1
ip netns exec net1 ip link set ipvlan1 up
ip netns exec net1 ip addr add 192.168.2.1/24 dev ipvlan1
ip route add 192.168.2.0/24 via 192.168.1.2
ping 192.168.2.2 -c 2
ip netns exec net1 ip link set ipvlan1 type ipvlan mode l2
ping 192.168.2.2 -c 2
Add the same configuration on remote host. After we set the mode to l2,
we could find that the src/dst MAC addresses are the same on eth1:
Fix this by calling dev_change_flags(), which will call netdevice notifier
with flag change info.
v2:
a) As pointed out by Wang Cong, check return value for dev_change_flags() when
change dev flags.
b) As suggested by Stefano and Sabrina, move flags setting before l3mdev_ops.
So we don't need to redo ipvlan_{, un}register_nf_hook() again in err path.
Reported-by: Jianlin Shi <jishi@redhat.com> Reviewed-by: Stefano Brivio <sbrivio@redhat.com> Reviewed-by: Sabrina Dubroca <sd@queasysnail.net> Fixes: 2ad7bf3638411 ("ipvlan: Initial check-in of the IPVLAN driver.") Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Biggers [Sat, 30 Jun 2018 22:26:56 +0000 (15:26 -0700)]
ipv6: sr: fix passing wrong flags to crypto_alloc_shash()
The 'mask' argument to crypto_alloc_shash() uses the CRYPTO_ALG_* flags,
not 'gfp_t'. So don't pass GFP_KERNEL to it.
Fixes: bf355b8d2c30 ("ipv6: sr: add core files for SR HMAC support") Signed-off-by: Eric Biggers <ebiggers@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Sabrina Dubroca [Sat, 30 Jun 2018 15:38:55 +0000 (17:38 +0200)]
net: fix use-after-free in GRO with ESP
Since the addition of GRO for ESP, gro_receive can consume the skb and
return -EINPROGRESS. In that case, the lower layer GRO handler cannot
touch the skb anymore.
Commit 5f114163f2f5 ("net: Add a skb_gro_flush_final helper.") converted
some of the gro_receive handlers that can lead to ESP's gro_receive so
that they wouldn't access the skb when -EINPROGRESS is returned, but
missed other spots, mainly in tunneling protocols.
This patch finishes the conversion to using skb_gro_flush_final(), and
adds a new helper, skb_gro_flush_final_remcsum(), used in VXLAN and
GUE.
Fixes: 5f114163f2f5 ("net: Add a skb_gro_flush_final helper.") Signed-off-by: Sabrina Dubroca <sd@queasysnail.net> Reviewed-by: Stefano Brivio <sbrivio@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Merge tag 'for-4.18-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
"We have a few regression fixes for qgroup rescan status tracking and
the vm_fault_t conversion that mixed up the error values"
* tag 'for-4.18-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
Btrfs: fix mount failure when qgroup rescan is in progress
Btrfs: fix regression in btrfs_page_mkwrite() from vm_fault_t conversion
btrfs: quota: Set rescan progress to (u64)-1 if we hit last leaf
Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull vfs fix from Al Viro:
"Followup to procfs-seq_file series this window"
This fixes a memory leak by making sure that proc seq files release any
private data on close. The 'proc_seq_open' has to be properly paired
with 'proc_seq_release' that releases the extra private data.
* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
proc: add proc_seq_release
Merge tag 'staging-4.18-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging
Pull staging/IIO fixes from Greg KH:
"Here are a few small staging and IIO driver fixes for 4.18-rc3.
Nothing major or big, all just fixes for reported problems since
4.18-rc1. All of these have been in linux-next this week with no
reported problems"
* tag 'staging-4.18-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
staging: android: ion: Return an ERR_PTR in ion_map_kernel
staging: comedi: quatech_daqp_cs: fix no-op loop daqp_ao_insn_write()
iio: imu: inv_mpu6050: Fix probe() failure on older ACPI based machines
iio: buffer: fix the function signature to match implementation
iio: mma8452: Fix ignoring MMA8452_INT_DRDY
iio: tsl2x7x/tsl2772: avoid potential division by zero
iio: pressure: bmp280: fix relative humidity unit
projects / mirror_ubuntu-disco-kernel.git / log