Denys provided an awesome KASAN report pointing to an use
after free in xt_TCPMSS
I have provided three patches to fix this issue, either in xt_TCPMSS or
in xt_tcpudp.c. It seems xt_TCPMSS patch has the smallest possible
impact.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Denys Fedoryshchenko <nuclearcat@nuclearcat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Markus Trippelsdorf reported that after commit dcb17d22e1 ("tcp: warn
on bogus MSS and try to amend it") the kernel started logging the
warning for a NIC driver that doesn't even support GRO.
It was diagnosed that it was possibly caused on connections that were
using TCP Timestamps but some packets lacked the Timestamps option. As
we reduce rcv_mss when timestamps are used, the lack of them would cause
the packets to be bigger than expected, although this is a valid case.
As this warning is more as a hint, getting a clean-cut on the
threshold is probably not worth the execution time spent on it. This
patch thus alleviates the false-positives with 2 quick checks: by
accounting for the entire TCP option space and also checking against the
interface MTU if it's available.
These changes, specially the MTU one, might mask some real positives,
though if they are really happening, it's possible that sooner or later
it will be triggered anyway.
Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is to move sctp_transport_dst_check into sctp_packet_config
from sctp_packet_transmit and add pathmtu check in sctp_packet_config.
With this fix, sctp can update dst or pathmtu before appending chunks,
which can void dropping packets in sctp_packet_transmit when dst is
obsolete or dst's mtu is changed.
This patch is also to improve some other codes in sctp_packet_config.
It updates packet max_size with gso_max_size, checks for dst and
pathmtu, and appends ecne chunk only when packet is empty and asoc
is not NULL.
It makes sctp flush work better, as we only need to set up them once
for one flush schedule. It's also safe, since asoc is NULL only when
the packet is created by sctp_ootb_pkt_new in which it just gets the
new dst, no need to do more things for it other than set packet with
transport's pathmtu.
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>
The last argument to __skb_header_pointer() should be a buffer large
enough to store struct arphdr. This can be a pointer to a struct arphdr
structure. The code was previously using a pointer to a pointer to
struct arphdr.
By my counting the storage available both before and after is 8 bytes on
x86_64.
Fixes: 55733350e5 ("flow disector: ARP support")
Reported-by: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In case, if TX watchdog is fired some or all netdev TX queues will be
stopped and as part of recovery it is required not only to drain and
reinitailize CPSW TX channeles, but also wake up stoppted TX queues what
doesn't happen now and netdevice will stop transmiting data until
reopenned.
Hence, add netif_tx_wake_all_queues() call in .ndo_tx_timeout() to complete
recovery and restore TX path.
Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Guillaume Nault says:
====================
l2tp: fix usage of l2tp_session_find()
l2tp_session_find() doesn't take a reference on the session returned to
its caller. Virtually all l2tp_session_find() users are racy, either
because the session can disappear from under them or because they take
a reference too late. This leads to bugs like 'use after free' or
failure to notice duplicate session creations.
In some cases, taking a reference on the session is not enough. The
special callbacks .ref() and .deref() also have to be called in cases
where the PPP pseudo-wire uses the socket associated with the session.
Therefore, when looking up a session, we also have to pass a flag
indicating if the .ref() callback has to be called.
In the future, we probably could drop the .ref() and .deref() callbacks
entirely by protecting the .sock field of struct pppol2tp_session with
RCU, thus allowing it to be freed and set to NULL even if the L2TP
session is still alive.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Callers of l2tp_nl_session_find() need to hold a reference on the
returned session since there's no guarantee that it isn't going to
disappear from under them.
Relying on the fact that no l2tp netlink message may be processed
concurrently isn't enough: sessions can be deleted by other means
(e.g. by closing the PPPOL2TP socket of a ppp pseudowire).
l2tp_nl_cmd_session_delete() is a bit special: it runs a callback
function that may require a previous call to session->ref(). In
particular, for ppp pseudowires, the callback is l2tp_session_delete(),
which then calls pppol2tp_session_close() and dereferences the PPPOL2TP
socket. The socket might already be gone at the moment
l2tp_session_delete() calls session->ref(), so we need to take a
reference during the session lookup. So we need to pass the do_ref
variable down to l2tp_session_get() and l2tp_session_get_by_ifname().
Since all callers have to be updated, l2tp_session_find_by_ifname() and
l2tp_nl_session_find() are renamed to reflect their new behaviour.
Fixes: 309795f4be ("l2tp: Add netlink control API for L2TP")
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
l2tp_session_find() doesn't take any reference on the returned session.
Therefore, the session may disappear while sending the notification.
Use l2tp_session_get() instead and decrement session's refcount once
the notification is sent.
Fixes: 33f72e6f0c ("l2tp : multicast notification to the registered listeners")
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
l2tp_session_create() relies on its caller for checking for duplicate
sessions. This is racy since a session can be concurrently inserted
after the caller's verification.
Fix this by letting l2tp_session_create() verify sessions uniqueness
upon insertion. Callers need to be adapted to check for
l2tp_session_create()'s return code instead of calling
l2tp_session_find().
pppol2tp_connect() is a bit special because it has to work on existing
sessions (if they're not connected) or to create a new session if none
is found. When acting on a preexisting session, a reference must be
held or it could go away on us. So we have to use l2tp_session_get()
instead of l2tp_session_find() and drop the reference before exiting.
Fixes: d9e31d17ce ("l2tp: Add L2TP ethernet pseudowire support")
Fixes: fd558d186d ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts")
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
Holding a reference on session is required before calling
pppol2tp_session_ioctl(). The session could get freed while processing the
ioctl otherwise. Since pppol2tp_session_ioctl() uses the session's socket,
we also need to take a reference on it in l2tp_session_get().
Fixes: fd558d186d ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts")
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
Taking a reference on sessions in l2tp_recv_common() is racy; this
has to be done by the callers.
To this end, a new function is required (l2tp_session_get()) to
atomically lookup a session and take a reference on it. Callers then
have to manually drop this reference.
Fixes: fd558d186d ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts")
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since sctp reconf was added in sctp, the real cnt of in/out stream
have not been c.sinit_max_instreams and c.sinit_num_ostreams any
more.
This patch is to replace them with stream->in/outcnt.
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>
Daniel Borkmann says:
====================
BPF fixes on map_value_adj reg types
This set adds two fixes for map_value_adj register type in the
verifier and user space tests along with them for the BPF self
test suite. For details, please see individual patches.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a couple of test cases, for example, probing for xadd on a spilled
pointer to packet and map_value_adj register, various other map_value_adj
tests including the unaligned load/store, and trying out pointer arithmetic
on map_value_adj register itself. For the unaligned load/store, we need
to figure out whether the architecture has efficient unaligned access and
need to mark affected tests accordingly.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, the verifier doesn't reject unaligned access for map_value_adj
register types. Commit 484611357c ("bpf: allow access into map value
arrays") added logic to check_ptr_alignment() extending it from PTR_TO_PACKET
to also PTR_TO_MAP_VALUE_ADJ, but for PTR_TO_MAP_VALUE_ADJ no enforcement
is in place, because reg->id for PTR_TO_MAP_VALUE_ADJ reg types is never
non-zero, meaning, we can cause BPF_H/_W/_DW-based unaligned access for
architectures not supporting efficient unaligned access, and thus worst
case could raise exceptions on some archs that are unable to correct the
unaligned access or perform a different memory access to the actual
requested one and such.
i) Unaligned load with !CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS
on r0 (map_value_adj):
0: (bf) r2 = r10
1: (07) r2 += -8
2: (7a) *(u64 *)(r2 +0) = 0
3: (18) r1 = 0x42533a00
5: (85) call bpf_map_lookup_elem#1
6: (15) if r0 == 0x0 goto pc+11
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
7: (61) r1 = *(u32 *)(r0 +0)
8: (35) if r1 >= 0xb goto pc+9
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R1=inv,min_value=0,max_value=10 R10=fp
9: (07) r0 += 3
10: (79) r7 = *(u64 *)(r0 +0)
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R1=inv,min_value=0,max_value=10 R10=fp
11: (79) r7 = *(u64 *)(r0 +2)
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R1=inv,min_value=0,max_value=10 R7=inv R10=fp
[...]
ii) Unaligned store with !CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS
on r0 (map_value_adj):
0: (bf) r2 = r10
1: (07) r2 += -8
2: (7a) *(u64 *)(r2 +0) = 0
3: (18) r1 = 0x4df16a00
5: (85) call bpf_map_lookup_elem#1
6: (15) if r0 == 0x0 goto pc+19
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
7: (07) r0 += 3
8: (7a) *(u64 *)(r0 +0) = 42
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R10=fp
9: (7a) *(u64 *)(r0 +2) = 43
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R10=fp
10: (7a) *(u64 *)(r0 -2) = 44
R0=map_value_adj(ks=8,vs=48,id=0),min_value=3,max_value=3 R10=fp
[...]
For the PTR_TO_PACKET type, reg->id is initially zero when skb->data
was fetched, it later receives a reg->id from env->id_gen generator
once another register with UNKNOWN_VALUE type was added to it via
check_packet_ptr_add(). The purpose of this reg->id is twofold: i) it
is used in find_good_pkt_pointers() for setting the allowed access
range for regs with PTR_TO_PACKET of same id once verifier matched
on data/data_end tests, and ii) for check_ptr_alignment() to determine
that when not having efficient unaligned access and register with
UNKNOWN_VALUE was added to PTR_TO_PACKET, that we're only allowed
to access the content bytewise due to unknown unalignment. reg->id
was never intended for PTR_TO_MAP_VALUE{,_ADJ} types and thus is
always zero, the only marking is in PTR_TO_MAP_VALUE_OR_NULL that
was added after 484611357c via 57a09bf0a4 ("bpf: Detect identical
PTR_TO_MAP_VALUE_OR_NULL registers"). Above tests will fail for
non-root environment due to prohibited pointer arithmetic.
The fix splits register-type specific checks into their own helper
instead of keeping them combined, so we don't run into a similar
issue in future once we extend check_ptr_alignment() further and
forget to add reg->type checks for some of the checks.
Fixes: 484611357c ("bpf: allow access into map value arrays")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
While looking into map_value_adj, I noticed that alu operations
directly on the map_value() resp. map_value_adj() register (any
alu operation on a map_value() register will turn it into a
map_value_adj() typed register) are not sufficiently protected
against some of the operations. Two non-exhaustive examples are
provided that the verifier needs to reject:
i) BPF_AND on r0 (map_value_adj):
0: (bf) r2 = r10
1: (07) r2 += -8
2: (7a) *(u64 *)(r2 +0) = 0
3: (18) r1 = 0xbf842a00
5: (85) call bpf_map_lookup_elem#1
6: (15) if r0 == 0x0 goto pc+2
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
7: (57) r0 &= 8
8: (7a) *(u64 *)(r0 +0) = 22
R0=map_value_adj(ks=8,vs=48,id=0),min_value=0,max_value=8 R10=fp
9: (95) exit
from 6 to 9: R0=inv,min_value=0,max_value=0 R10=fp
9: (95) exit
processed 10 insns
ii) BPF_ADD in 32 bit mode on r0 (map_value_adj):
0: (bf) r2 = r10
1: (07) r2 += -8
2: (7a) *(u64 *)(r2 +0) = 0
3: (18) r1 = 0xc24eee00
5: (85) call bpf_map_lookup_elem#1
6: (15) if r0 == 0x0 goto pc+2
R0=map_value(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
7: (04) (u32) r0 += (u32) 0
8: (7a) *(u64 *)(r0 +0) = 22
R0=map_value_adj(ks=8,vs=48,id=0),min_value=0,max_value=0 R10=fp
9: (95) exit
from 6 to 9: R0=inv,min_value=0,max_value=0 R10=fp
9: (95) exit
processed 10 insns
Issue is, while min_value / max_value boundaries for the access
are adjusted appropriately, we change the pointer value in a way
that cannot be sufficiently tracked anymore from its origin.
Operations like BPF_{AND,OR,DIV,MUL,etc} on a destination register
that is PTR_TO_MAP_VALUE{,_ADJ} was probably unintended, in fact,
all the test cases coming with 484611357c ("bpf: allow access
into map value arrays") perform BPF_ADD only on the destination
register that is PTR_TO_MAP_VALUE_ADJ.
Only for UNKNOWN_VALUE register types such operations make sense,
f.e. with unknown memory content fetched initially from a constant
offset from the map value memory into a register. That register is
then later tested against lower / upper bounds, so that the verifier
can then do the tracking of min_value / max_value, and properly
check once that UNKNOWN_VALUE register is added to the destination
register with type PTR_TO_MAP_VALUE{,_ADJ}. This is also what the
original use-case is solving. Note, tracking on what is being
added is done through adjust_reg_min_max_vals() and later access
to the map value enforced with these boundaries and the given offset
from the insn through check_map_access_adj().
Tests will fail for non-root environment due to prohibited pointer
arithmetic, in particular in check_alu_op(), we bail out on the
is_pointer_value() check on the dst_reg (which is false in root
case as we allow for pointer arithmetic via env->allow_ptr_leaks).
Similarly to PTR_TO_PACKET, one way to fix it is to restrict the
allowed operations on PTR_TO_MAP_VALUE{,_ADJ} registers to 64 bit
mode BPF_ADD. The test_verifier suite runs fine after the patch
and it also rejects mentioned test cases.
Fixes: 484611357c ("bpf: allow access into map value arrays")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Without this the generic cdc_ether grabs the device,
and does not really work.
Signed-off-by: René Rebe <rene@exactcode.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
ovs_flow_key_update() is called when the flow key is invalid, and it is
used to update and revalidate the flow key. Commit 329f45bc4f
("openvswitch: add mac_proto field to the flow key") introduces mac_proto
field to flow key and use it to determine whether the flow key is valid.
However, the commit does not update the code path in ovs_flow_key_update()
to revalidate the flow key which may cause BUG_ON() on execute_recirc().
This patch addresses the aforementioned issue.
Fixes: 329f45bc4f ("openvswitch: add mac_proto field to the flow key")
Signed-off-by: Yi-Hung Wei <yihung.wei@gmail.com>
Acked-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This driver uses interfaces from linux/of.h and linux/property.h but
relies on implict inclusion of those headers which means that changes in
other headers could break the build, as happened in -next for arm today.
Add a explicit includes.
Signed-off-by: Mark Brown <broonie@kernel.org>
Acked-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
The current code only supports DT to check SFP present.
This patch adds ACPI support as well.
Signed-off-by: Daode Huang <huangdaode@hisilicon.com>
Reviewed-by: Yisen Zhuang <yisen.zhuang@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use cpu_to_le32() for link_config variable in set_logical_link_config
command as this variable is of type u32.
Signed-off-by: Suresh Reddy <suresh.reddy@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When sending a msg without asoc established, sctp will send INIT packet
first and then enqueue chunks.
Before receiving INIT_ACK, stream info is not yet alloced. But enqueuing
chunks needs to access stream info, like out stream state and out stream
cnt.
This patch is to fix it by allocing out stream info when initializing an
asoc, allocing in stream and re-allocing out stream when processing init.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When calculating po->tp_hdrlen + po->tp_reserve the result can overflow.
Fix by checking that tp_reserve <= INT_MAX on assign.
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When calculating rb->frames_per_block * req->tp_block_nr the result
can overflow.
Add a check that tp_block_size * tp_block_nr <= UINT_MAX.
Since frames_per_block <= tp_block_size, the expression would
never overflow.
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Subtracting tp_sizeof_priv from tp_block_size and casting to int
to check whether one is less then the other doesn't always work
(both of them are unsigned ints).
Compare them as is instead.
Also cast tp_sizeof_priv to u64 before using BLK_PLUS_PRIV, as
it can overflow inside BLK_PLUS_PRIV otherwise.
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pablo Neira Ayuso says:
====================
Netfilter fixes for net
The following patchset contains a rather large update with Netfilter
fixes, specifically targeted to incorrect RCU usage in several spots and
the userspace conntrack helper infrastructure (nfnetlink_cthelper),
more specifically they are:
1) expect_class_max is incorrect set via cthelper, as in kernel semantics
mandate that this represents the array of expectation classes minus 1.
Patch from Liping Zhang.
2) Expectation policy updates via cthelper are currently broken for several
reasons: This code allows illegal changes in the policy such as changing
the number of expeciation classes, it is leaking the updated policy and
such update occurs with no RCU protection at all. Fix this by adding a
new nfnl_cthelper_update_policy() that describes what is really legal on
the update path.
3) Fix several memory leaks in cthelper, from Jeffy Chen.
4) synchronize_rcu() is missing in the removal path of several modules,
this may lead to races since CPU may still be running on code that has
just gone. Also from Liping Zhang.
5) Don't use the helper hashtable from cthelper, it is not safe to walk
over those bits without the helper mutex. Fix this by introducing a
new independent list for userspace helpers. From Liping Zhang.
6) nf_ct_extend_unregister() needs synchronize_rcu() to make sure no
packets are walking on any conntrack extension that is gone after
module removal, again from Liping.
7) nf_nat_snmp may crash if we fail to unregister the helper due to
accidental leftover code, from Gao Feng.
8) Fix leak in nfnetlink_queue with secctx support, from Liping Zhang.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
After a new NAPI_STATE_MISSED state was added to NAPI we can get into
this state and in such case we have to reschedule NAPI as some work is
still pending and we have to process it. napi_complete_done() function
returns false if we have to reschedule something (e.g. in case we were
in MISSED state) as current polling have not been completed yet.
nps_enet driver hasn't been verifying the return value of
napi_complete_done() and has been forcibly enabling interrupts. That is
not correct as we should not enable interrupts before we have processed
all scheduled work. As a result we were getting trapped in interrupt
hanlder chain as we had never been able to disabale ethernet
interrupts again.
So this patch makes nps_enet_poll() func verify return value of
napi_complete_done() and enable interrupts only in case all scheduled
work has been completed.
Signed-off-by: Vlad Zakharov <vzakhar@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan says:
====================
bnxt_en: Small misc. fixes.
Fix a NULL pointer crash in open failure path, wrong arguments when
printing error messages, and a DMA unmap bug in XDP shutdown path.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
In bnxt_free_rx_skbs(), which is called to free up all RX buffers during
shutdown, we need to unmap the page if we are running in XDP mode.
Fixes: c61fb99cae ("bnxt_en: Add RX page mode support.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sankar Patchineelam <sankar.patchineelam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Net device reset can fail when the h/w or f/w is in a bad state.
Subsequent netdevice open fails in bnxt_hwrm_stat_ctx_alloc().
The cleanup invokes bnxt_hwrm_resource_free() which inturn
calls bnxt_disable_int(). In this routine, the code segment
if (ring->fw_ring_id != INVALID_HW_RING_ID)
BNXT_CP_DB(cpr->cp_doorbell, cpr->cp_raw_cons);
results in NULL pointer dereference as cpr->cp_doorbell is not yet
initialized, and fw_ring_id is zero.
The fix is to initialize cpr fw_ring_id to INVALID_HW_RING_ID before
bnxt_init_chip() is invoked.
Signed-off-by: Sankar Patchineelam <sankar.patchineelam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The Rx path may grab the socket right before pppol2tp_release(), but
nothing guarantees that it will enqueue packets before
skb_queue_purge(). Therefore, the socket can be destroyed without its
queues fully purged.
Fix this by purging queues in pppol2tp_session_destruct() where we're
guaranteed nothing is still referencing the socket.
Fixes: 9e9cb6221a ("l2tp: fix userspace reception on plain L2TP sockets")
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
The code following l2tp_tunnel_find() expects that a new reference is
held on sk. Either sk_receive_skb() or the discard_put error path will
drop a reference from the tunnel's socket.
This issue exists in both l2tp_ip and l2tp_ip6.
Fixes: a3c18422a4 ("l2tp: hold socket before dropping lock in l2tp_ip{, 6}_recv()")
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
When internal mac80211 TXQs aren't supported, netdev queues must
always started out started even when driver queues are stopped
while the interface is added. This is necessary because with the
internal TXQ support netdev queues are never stopped and packet
scheduling/dropping is done in mac80211.
Cc: stable@vger.kernel.org # 4.9+
Fixes: 80a83cfc43 ("mac80211: skip netdev queue control with software queuing")
Reported-and-tested-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
We must call security_release_secctx to free the memory returned by
security_secid_to_secctx, otherwise memory may be leaked forever.
Fixes: ef493bd930 ("netfilter: nfnetlink_queue: add security context information")
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
We got the following use-after-free KASAN report:
BUG: KASAN: use-after-free in wiphy_resume+0x591/0x5a0 [cfg80211]
at addr ffff8803fc244090
Read of size 8 by task kworker/u16:24/2587
CPU: 6 PID: 2587 Comm: kworker/u16:24 Tainted: G B 4.9.13-debug+
Hardware name: Dell Inc. XPS 15 9550/0N7TVV, BIOS 1.2.19 12/22/2016
Workqueue: events_unbound async_run_entry_fn
ffff880425d4f9d8 ffffffffaeedb541 ffff88042b80ef00 ffff8803fc244088
ffff880425d4fa00 ffffffffae84d7a1 ffff880425d4fa98 ffff8803fc244080
ffff88042b80ef00 ffff880425d4fa88 ffffffffae84da3a ffffffffc141f7d9
Call Trace:
[<ffffffffaeedb541>] dump_stack+0x85/0xc4
[<ffffffffae84d7a1>] kasan_object_err+0x21/0x70
[<ffffffffae84da3a>] kasan_report_error+0x1fa/0x500
[<ffffffffc141f7d9>] ? cfg80211_bss_age+0x39/0xc0 [cfg80211]
[<ffffffffc141f83a>] ? cfg80211_bss_age+0x9a/0xc0 [cfg80211]
[<ffffffffae48d46d>] ? trace_hardirqs_on+0xd/0x10
[<ffffffffc13fb1c0>] ? wiphy_suspend+0xc70/0xc70 [cfg80211]
[<ffffffffae84def1>] __asan_report_load8_noabort+0x61/0x70
[<ffffffffc13fb100>] ? wiphy_suspend+0xbb0/0xc70 [cfg80211]
[<ffffffffc13fb751>] ? wiphy_resume+0x591/0x5a0 [cfg80211]
[<ffffffffc13fb751>] wiphy_resume+0x591/0x5a0 [cfg80211]
[<ffffffffc13fb1c0>] ? wiphy_suspend+0xc70/0xc70 [cfg80211]
[<ffffffffaf3b206e>] dpm_run_callback+0x6e/0x4f0
[<ffffffffaf3b31b2>] device_resume+0x1c2/0x670
[<ffffffffaf3b367d>] async_resume+0x1d/0x50
[<ffffffffae3ee84e>] async_run_entry_fn+0xfe/0x610
[<ffffffffae3d0666>] process_one_work+0x716/0x1a50
[<ffffffffae3d05c9>] ? process_one_work+0x679/0x1a50
[<ffffffffafdd7b6d>] ? _raw_spin_unlock_irq+0x3d/0x60
[<ffffffffae3cff50>] ? pwq_dec_nr_in_flight+0x2b0/0x2b0
[<ffffffffae3d1a80>] worker_thread+0xe0/0x1460
[<ffffffffae3d19a0>] ? process_one_work+0x1a50/0x1a50
[<ffffffffae3e54c2>] kthread+0x222/0x2e0
[<ffffffffae3e52a0>] ? kthread_park+0x80/0x80
[<ffffffffae3e52a0>] ? kthread_park+0x80/0x80
[<ffffffffae3e52a0>] ? kthread_park+0x80/0x80
[<ffffffffafdd86aa>] ret_from_fork+0x2a/0x40
Object at ffff8803fc244088, in cache kmalloc-1024 size: 1024
Allocated:
PID = 71
save_stack_trace+0x1b/0x20
save_stack+0x46/0xd0
kasan_kmalloc+0xad/0xe0
kasan_slab_alloc+0x12/0x20
__kmalloc_track_caller+0x134/0x360
kmemdup+0x20/0x50
brcmf_cfg80211_attach+0x10b/0x3a90 [brcmfmac]
brcmf_bus_start+0x19a/0x9a0 [brcmfmac]
brcmf_pcie_setup+0x1f1a/0x3680 [brcmfmac]
brcmf_fw_request_nvram_done+0x44c/0x11b0 [brcmfmac]
request_firmware_work_func+0x135/0x280
process_one_work+0x716/0x1a50
worker_thread+0xe0/0x1460
kthread+0x222/0x2e0
ret_from_fork+0x2a/0x40
Freed:
PID = 2568
save_stack_trace+0x1b/0x20
save_stack+0x46/0xd0
kasan_slab_free+0x71/0xb0
kfree+0xe8/0x2e0
brcmf_cfg80211_detach+0x62/0xf0 [brcmfmac]
brcmf_detach+0x14a/0x2b0 [brcmfmac]
brcmf_pcie_remove+0x140/0x5d0 [brcmfmac]
brcmf_pcie_pm_leave_D3+0x198/0x2e0 [brcmfmac]
pci_pm_resume+0x186/0x220
dpm_run_callback+0x6e/0x4f0
device_resume+0x1c2/0x670
async_resume+0x1d/0x50
async_run_entry_fn+0xfe/0x610
process_one_work+0x716/0x1a50
worker_thread+0xe0/0x1460
kthread+0x222/0x2e0
ret_from_fork+0x2a/0x40
Memory state around the buggy address:
ffff8803fc243f80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
ffff8803fc244000: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>ffff8803fc244080: fc fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
^
ffff8803fc244100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff8803fc244180: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
What is happening is that brcmf_pcie_resume() detects a device that
is no longer responsive and it decides to unbind resulting in a
wiphy_unregister() and wiphy_free() call. Now the wiphy instance
remains allocated, because PM needs to call wiphy_resume() for it.
However, brcmfmac already does a kfree() for the struct
cfg80211_registered_device::ops field. Change the checks in
wiphy_resume() to only access the struct cfg80211_registered_device::ops
if the wiphy instance is still registered at this time.
Cc: stable@vger.kernel.org # 4.10.x, 4.9.x
Reported-by: Daniel J Blueman <daniel@quora.org>
Reviewed-by: Hante Meuleman <hante.meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieter-paul.giesberts@broadcom.com>
Reviewed-by: Franky Lin <franky.lin@broadcom.com>
Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The reference count held for skb needs to be released when the skb's
nfct pointer is cleared regardless of if nf_ct_delete() is called or
not.
Failing to release the skb's reference cound led to deferred conntrack
cleanup spinning forever within nf_conntrack_cleanup_net_list() when
cleaning up a network namespace:
kworker/u16:0-19025 [004] 45981067.173642: sched_switch: kworker/u16:0:19025 [120] R ==> rcu_preempt:7 [120]
kworker/u16:0-19025 [004] 45981067.173651: kernel_stack: <stack trace>
=> ___preempt_schedule (ffffffffa001ed36)
=> _raw_spin_unlock_bh (ffffffffa0713290)
=> nf_ct_iterate_cleanup (ffffffffc00a4454)
=> nf_conntrack_cleanup_net_list (ffffffffc00a5e1e)
=> nf_conntrack_pernet_exit (ffffffffc00a63dd)
=> ops_exit_list.isra.1 (ffffffffa06075f3)
=> cleanup_net (ffffffffa0607df0)
=> process_one_work (ffffffffa0084c31)
=> worker_thread (ffffffffa008592b)
=> kthread (ffffffffa008bee2)
=> ret_from_fork (ffffffffa071b67c)
Fixes: dd41d33f0b ("openvswitch: Add force commit.")
Reported-by: Yang Song <yangsong@vmware.com>
Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
Acked-by: Joe Stringer <joe@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
gcc-7 reports a warning that earlier versions did not have:
drivers/net/ethernet/rocker/rocker_ofdpa.c: In function 'ofdpa_port_stp_update':
arch/x86/include/asm/string_32.h:79:22: error: '*((void *)&prev_ctrls+4)' may be used uninitialized in this function [-Werror=maybe-uninitialized]
*((short *)to + 2) = *((short *)from + 2);
~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~
drivers/net/ethernet/rocker/rocker_ofdpa.c:2218:7: note: '*((void *)&prev_ctrls+4)' was declared here
This is clearly a variation of the warning about 'prev_state' that
was shut up using uninitialized_var().
We can slightly simplify the code and get rid of the warning by unconditionally
saving the prev_state and prev_ctrls variables. The inlined memcpy is not
particularly expensive here, as it just has to read five bytes from one or
two cache lines.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
In NETDEV_CHANGEUPPER event the upper_info field is valid
only when linking is true. Otherwise it should be ignored.
Fixes: 7907f23adc (net/mlx5: Implement RoCE LAG feature)
Signed-off-by: Talat Batheesh <talatb@mellanox.com>
Reviewed-by: Aviv Heller <avivh@mellanox.com>
Reviewed-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
moxart_mac_start_xmit() doesn't care where tx_tail is, tx_head can
catch and pass tx_tail, which is bad because moxart_tx_finished()
isn't guaranteed to catch up on freeing resources from tx_tail.
Add a check in moxart_mac_start_xmit() stopping the queue at the
end of the circular buffer. Also add a check in moxart_tx_finished()
waking the queue if the buffer has TX_WAKE_THRESHOLD or more
free descriptors.
While we're at it, move spin_lock_irq() to happen before our
descriptor pointer is assigned in moxart_mac_start_xmit().
Addresses https://bugzilla.kernel.org/show_bug.cgi?id=99451
Signed-off-by: Jonas Jensen <jonas.jensen@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
gcc-7 points out that the AVMB1_ADDCARD ioctl results in an unintialized
value ending up in the cardnr parameter:
drivers/isdn/capi/kcapi.c: In function 'old_capi_manufacturer':
drivers/isdn/capi/kcapi.c:1042:24: error: 'cdef.cardnr' may be used uninitialized in this function [-Werror=maybe-uninitialized]
cparams.cardnr = cdef.cardnr;
This has been broken since before the start of the git history, so
either the value is not used for anything important, or the ioctl
command doesn't get called in practice.
Setting the cardnr to zero avoids the warning and makes sure
we have consistent behavior.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
David Laight noticed the support for MSG_MORE with datamsg->force_delay
didn't really work as we expected, as the first msg with MSG_MORE set
would always block the following chunks' dequeuing.
This Patch is to rewrite it by saving the MSG_MORE flag into assoc as
David Laight suggested.
asoc->force_delay is used to save MSG_MORE flag before a msg is sent.
All chunks in queue would not be sent out if asoc->force_delay is set
by the msg with MSG_MORE flag, until a new msg without MSG_MORE flag
clears asoc->force_delay.
Note that this change would not affect the flush is generated by other
triggers, like asoc->state != ESTABLISHED, queue size > pmtu etc.
v1->v2:
Not clear asoc->force_delay after sending the msg with MSG_MORE flag.
Fixes: 4ea0c32f5f ("sctp: add support for MSG_MORE")
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: David Laight <david.laight@aculab.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Our chosen ic_dev may be anywhere in our list of ic_devs, and we may
free it before attempting to close others. When we compare d->dev and
ic_dev->dev, we're potentially dereferencing memory returned to the
allocator. This causes KASAN to scream for each subsequent ic_dev we
check.
As there's a 1-1 mapping between ic_devs and netdevs, we can instead
compare d and ic_dev directly, which implicitly handles the !ic_dev
case, and avoids the use-after-free. The ic_dev pointer may be stale,
but we will not dereference it.
Original splat:
[ 6.487446] ==================================================================
[ 6.494693] BUG: KASAN: use-after-free in ic_close_devs+0xc4/0x154 at addr ffff800367efa708
[ 6.503013] Read of size 8 by task swapper/0/1
[ 6.507452] CPU: 5 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc3-00002-gda42158 #8
[ 6.514993] Hardware name: AppliedMicro Mustang/Mustang, BIOS 3.05.05-beta_rc Jan 27 2016
[ 6.523138] Call trace:
[ 6.525590] [<ffff200008094778>] dump_backtrace+0x0/0x570
[ 6.530976] [<ffff200008094d08>] show_stack+0x20/0x30
[ 6.536017] [<ffff200008bee928>] dump_stack+0x120/0x188
[ 6.541231] [<ffff20000856d5e4>] kasan_object_err+0x24/0xa0
[ 6.546790] [<ffff20000856d924>] kasan_report_error+0x244/0x738
[ 6.552695] [<ffff20000856dfec>] __asan_report_load8_noabort+0x54/0x80
[ 6.559204] [<ffff20000aae86ac>] ic_close_devs+0xc4/0x154
[ 6.564590] [<ffff20000aaedbac>] ip_auto_config+0x2ed4/0x2f1c
[ 6.570321] [<ffff200008084b04>] do_one_initcall+0xcc/0x370
[ 6.575882] [<ffff20000aa31de8>] kernel_init_freeable+0x5f8/0x6c4
[ 6.581959] [<ffff20000a16df00>] kernel_init+0x18/0x190
[ 6.587171] [<ffff200008084710>] ret_from_fork+0x10/0x40
[ 6.592468] Object at ffff800367efa700, in cache kmalloc-128 size: 128
[ 6.598969] Allocated:
[ 6.601324] PID = 1
[ 6.603427] save_stack_trace_tsk+0x0/0x418
[ 6.607603] save_stack_trace+0x20/0x30
[ 6.611430] kasan_kmalloc+0xd8/0x188
[ 6.615087] ip_auto_config+0x8c4/0x2f1c
[ 6.619002] do_one_initcall+0xcc/0x370
[ 6.622832] kernel_init_freeable+0x5f8/0x6c4
[ 6.627178] kernel_init+0x18/0x190
[ 6.630660] ret_from_fork+0x10/0x40
[ 6.634223] Freed:
[ 6.636233] PID = 1
[ 6.638334] save_stack_trace_tsk+0x0/0x418
[ 6.642510] save_stack_trace+0x20/0x30
[ 6.646337] kasan_slab_free+0x88/0x178
[ 6.650167] kfree+0xb8/0x478
[ 6.653131] ic_close_devs+0x130/0x154
[ 6.656875] ip_auto_config+0x2ed4/0x2f1c
[ 6.660875] do_one_initcall+0xcc/0x370
[ 6.664705] kernel_init_freeable+0x5f8/0x6c4
[ 6.669051] kernel_init+0x18/0x190
[ 6.672534] ret_from_fork+0x10/0x40
[ 6.676098] Memory state around the buggy address:
[ 6.680880] ffff800367efa600: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[ 6.688078] ffff800367efa680: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[ 6.695276] >ffff800367efa700: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 6.702469] ^
[ 6.705952] ffff800367efa780: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[ 6.713149] ffff800367efa800: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 6.720343] ==================================================================
[ 6.727536] Disabling lock debugging due to kernel taint
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Cc: David S. Miller <davem@davemloft.net>
Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
Cc: James Morris <jmorris@namei.org>
Cc: Patrick McHardy <kaber@trash.net>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
Andrew has been contributing a lot to PHYLIB over the past months and
his feedback on patches is more than welcome.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the commit 93557f53e1 ("netfilter: nf_conntrack: nf_conntrack snmp
helper"), the snmp_helper is replaced by nf_nat_snmp_hook. So the
snmp_helper is never registered. But it still tries to unregister the
snmp_helper, it could cause the panic.
Now remove the useless snmp_helper and the unregister call in the
error handler.
Fixes: 93557f53e1 ("netfilter: nf_conntrack: nf_conntrack snmp helper")
Signed-off-by: Gao Feng <fgao@ikuai8.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
If one cpu is doing nf_ct_extend_unregister while another cpu is doing
__nf_ct_ext_add_length, then we may hit BUG_ON(t == NULL). Moreover,
there's no synchronize_rcu invocation after set nf_ct_ext_types[id] to
NULL, so it's possible that we may access invalid pointer.
But actually, most of the ct extends are built-in, so the problem listed
above will not happen. However, there are two exceptions: NF_CT_EXT_NAT
and NF_CT_EXT_SYNPROXY.
For _EXT_NAT, the panic will not happen, since adding the nat extend and
unregistering the nat extend are located in the same file(nf_nat_core.c),
this means that after the nat module is removed, we cannot add the nat
extend too.
For _EXT_SYNPROXY, synproxy extend may be added by init_conntrack, while
synproxy extend unregister will be done by synproxy_core_exit. So after
nf_synproxy_core.ko is removed, we may still try to add the synproxy
extend, then kernel panic may happen.
I know it's very hard to reproduce this issue, but I can play a tricky
game to make it happen very easily :)
Step 1. Enable SYNPROXY for tcp dport 1234 at FORWARD hook:
# iptables -I FORWARD -p tcp --dport 1234 -j SYNPROXY
Step 2. Queue the syn packet to the userspace at raw table OUTPUT hook.
Also note, in the userspace we only add a 20s' delay, then
reinject the syn packet to the kernel:
# iptables -t raw -I OUTPUT -p tcp --syn -j NFQUEUE --queue-num 1
Step 3. Using "nc 2.2.2.2 1234" to connect the server.
Step 4. Now remove the nf_synproxy_core.ko quickly:
# iptables -F FORWARD
# rmmod ipt_SYNPROXY
# rmmod nf_synproxy_core
Step 5. After 20s' delay, the syn packet is reinjected to the kernel.
Now you will see the panic like this:
kernel BUG at net/netfilter/nf_conntrack_extend.c:91!
Call Trace:
? __nf_ct_ext_add_length+0x53/0x3c0 [nf_conntrack]
init_conntrack+0x12b/0x600 [nf_conntrack]
nf_conntrack_in+0x4cc/0x580 [nf_conntrack]
ipv4_conntrack_local+0x48/0x50 [nf_conntrack_ipv4]
nf_reinject+0x104/0x270
nfqnl_recv_verdict+0x3e1/0x5f9 [nfnetlink_queue]
? nfqnl_recv_verdict+0x5/0x5f9 [nfnetlink_queue]
? nla_parse+0xa0/0x100
nfnetlink_rcv_msg+0x175/0x6a9 [nfnetlink]
[...]
One possible solution is to make NF_CT_EXT_SYNPROXY extend built-in, i.e.
introduce nf_conntrack_synproxy.c and only do ct extend register and
unregister in it, similar to nf_conntrack_timeout.c.
But having such a obscure restriction of nf_ct_extend_unregister is not a
good idea, so we should invoke synchronize_rcu after set nf_ct_ext_types
to NULL, and check the NULL pointer when do __nf_ct_ext_add_length. Then
it will be easier if we add new ct extend in the future.
Last, we use kfree_rcu to free nf_ct_ext, so rcu_barrier() is unnecessary
anymore, remove it too.
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Acked-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The nf_ct_helper_hash table is protected by nf_ct_helper_mutex, while
nfct_helper operation is protected by nfnl_lock(NFNL_SUBSYS_CTHELPER).
So it's possible that one CPU is walking the nf_ct_helper_hash for
cthelper add/get/del, another cpu is doing nf_conntrack_helpers_unregister
at the same time. This is dangrous, and may cause use after free error.
Note, delete operation will flush all cthelpers added via nfnetlink, so
using rcu to do protect is not easy.
Now introduce a dummy list to record all the cthelpers added via
nfnetlink, then we can walk the dummy list instead of walking the
nf_ct_helper_hash. Also, keep nfnl_cthelper_dump_table unchanged, it
may be invoked without nfnl_lock(NFNL_SUBSYS_CTHELPER) held.
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Otherwise, another CPU may access the invalid pointer. For example:
CPU0 CPU1
- rcu_read_lock();
- pfunc = _hook_;
_hook_ = NULL; -
mod unload -
- pfunc(); // invalid, panic
- rcu_read_unlock();
So we must call synchronize_rcu() to wait the rcu reader to finish.
Also note, in nf_nat_snmp_basic_fini, synchronize_rcu() will be invoked
by later nf_conntrack_helper_unregister, but I'm inclined to add a
explicit synchronize_rcu after set the nf_nat_snmp_hook to NULL. Depend
on such obscure assumptions is not a good idea.
Last, in nfnetlink_cttimeout, we use kfree_rcu to free the time object,
so in cttimeout_exit, invoking rcu_barrier() is not necessary at all,
remove it too.
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>