Currently in nf_conntrack_hash_check_insert(), when it fails in
nf_ct_ext_valid_pre/post(), NF_CT_STAT_INC() will be called in the
preemptible context, a call trace can be triggered:
BUG: using __this_cpu_add() in preemptible [00000000] code: conntrack/1636
caller is nf_conntrack_hash_check_insert+0x45/0x430 [nf_conntrack]
Call Trace:
<TASK>
dump_stack_lvl+0x33/0x46
check_preemption_disabled+0xc3/0xf0
nf_conntrack_hash_check_insert+0x45/0x430 [nf_conntrack]
ctnetlink_create_conntrack+0x3cd/0x4e0 [nf_conntrack_netlink]
ctnetlink_new_conntrack+0x1c0/0x450 [nf_conntrack_netlink]
nfnetlink_rcv_msg+0x277/0x2f0 [nfnetlink]
netlink_rcv_skb+0x50/0x100
nfnetlink_rcv+0x65/0x144 [nfnetlink]
netlink_unicast+0x1ae/0x290
netlink_sendmsg+0x257/0x4f0
sock_sendmsg+0x5f/0x70
This patch is to fix it by changing to use NF_CT_STAT_INC_ATOMIC() for
nf_ct_ext_valid_pre/post() check in nf_conntrack_hash_check_insert(),
as well as nf_ct_ext_valid_post() in __nf_conntrack_confirm().
Note that nf_ct_ext_valid_pre() check in __nf_conntrack_confirm() is
safe to use NF_CT_STAT_INC(), as it's under local_bh_disable().
Fixes: c56716c69c ("netfilter: extensions: introduce extension genid count")
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Change IPsec initialization flow to allow future creation of hardware
resources that should be released and allocated during devlink reload
operation. As part of that change, update function signature to be
void as no callers are actually interested in it.
Reviewed-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
TC trap action offload is currently supported only when trap is the sole action
in the flow.
This patch remove this limitation by changing trap action offload to not use
MLX5_ATTR_FLAG_SLOW_PATH flag and instead set the flow destination table
explicitly to be the slow table. This will allow offload of the additional
actions.
TC flow example:
tc filter add dev $REP2 protocol ip prio 2 root \
flower skip_sw dst_mac $mac0 \
action mirred egress redirect dev $REP3 \
action pedit ex munge eth dst set $mac2 pipe \
action trap
Signed-off-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Raed Salem <raeds@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Adding flow flag cases in setup vport dests before the slow path
case is incorrect as the slow path should take precedence.
Current code doesn't show this importance so make the slow path
case return early and separate from the other cases and remove
the redundant comparison of it in the sample case.
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Chris Mi <cmi@nvidia.com>
Reviewed-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
If ASO failed in creation, it won't be called to destroy either.
The kernel coding pattern is to make sure that callers are calling
to destroy only for valid objects.
Reviewed-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Current code used termination table for each vport destination
while it's only required for hairpin, i.e. uplink to uplink, or
when vlan push on rx action being used.
Fix to skip using termination table for vport destinations that
do not require it.
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
MLX5_UMR_KLM_ALIGNMENT is in units of number of entries, while
MLX5_UMR_MTT_ALIGNMENT (generalized and renamed to
MLX5_UMR_FLEX_ALIGNMENT) is in byte units. This is misleading and
confusing.
Replace this KLM definition with one based on the generic definition.
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Per the device spec, MLX5_UMR_MTT_ALIGNMENT is good not only for UMR MTT
entries, but for all other entries as well, like KLMs and KSMs.
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Defines MLX5_UMR_MTT_MASK and MLX5_UMR_MTT_MIN_CHUNK_SIZE are not in
use. Remove them.
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Per the device spec, MTTs/KLMs list in a UMR WQE must be aligned to 64B.
Per our SW design, the MTT/KLMs list would need alignment only if it's
too small, for example on PPC when PAGE_SIZE is 64KB, and only 4 pages
are needed to cover a MPWQE of size 256KB.
Padding, if needed, is taken into account when calculating the UMR WQE
fields (ds_cnt and xlt_octowords), however no entries are provided,
instead garbage is passed.
No real harm though, as these parts act as gaps between the RX MPWQEs
and not used by any of them. Hence, in practice, device does not try to
write any incoming packet to them. Still, prefer providing clean padding
marking the end of the list, and do not map garbage into the RQ memory
region.
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Remove mlx5_priv.ctx_list and ctx_lock which are no longer used after
commit 601c10c89c ("net/mlx5: Delete custom device management logic").
Signed-off-by: Petr Pavlu <petr.pavlu@suse.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
The mlx5 net files don't use io_mapping functionalities. So there is no
point in including <linux/io-mapping.h>.
Remove it.
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Steffen Klassert says:
====================
ipsec-next 2022-11-26
1) Remove redundant variable in esp6.
From Colin Ian King.
2) Update x->lastused for every packet. It was used only for
outgoing mobile IPv6 packets, but showed to be usefull
to check if the a SA is still in use in general.
From Antony Antony.
3) Remove unused variable in xfrm_byidx_resize.
From Leon Romanovsky.
4) Finalize extack support for xfrm.
From Sabrina Dubroca.
* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next:
xfrm: add extack to xfrm_set_spdinfo
xfrm: add extack to xfrm_alloc_userspi
xfrm: add extack to xfrm_do_migrate
xfrm: add extack to xfrm_new_ae and xfrm_replay_verify_len
xfrm: add extack to xfrm_del_sa
xfrm: add extack to xfrm_add_sa_expire
xfrm: a few coding style clean ups
xfrm: Remove not-used total variable
xfrm: update x->lastused for every packet
esp6: remove redundant variable err
====================
Link: https://lore.kernel.org/r/20221126110303.1859238-1-steffen.klassert@secunet.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Maxime Chevallier says:
====================
net: pcs: altera-tse: simplify and clean-up the driver
This small series does a bit of code cleanup in the altera TSE pcs
driver, removing unused register definitions, handling 1000BaseX speed
configuration correctly according to the datasheet, and making use of
proper poll_timeout helpers.
No functional change is introduced.
====================
Link: https://lore.kernel.org/r/20221125131801.64234-1-maxime.chevallier@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
remove unused register definitions, left from the split with the
altera-tse mac driver.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When disabling the SGMII mode bit, the PCS defaults to 1000BaseX mode.
In that mode, we don't need to set the speed since it's always 1000Mbps.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Software resets on the TSE PCS don't clear registers, but rather reset
all internal state machines regarding AN, comma detection and
encoding/decoding. Use read_poll_timeout to wait for the reset to clear
instead of manually polling the register.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Matthieu Baerts says:
====================
mptcp: MSG_FASTOPEN and TFO listener side support
Before this series, only the initiator of a connection was able to combine
both TCP FastOpen and MPTCP when using TCP_FASTOPEN_CONNECT socket option.
These new patches here add (in theory) the full support of TFO with MPTCP,
which means:
- MSG_FASTOPEN sendmsg flag support (patch 1/8)
- TFO support for the listener side (patches 2-5/8)
- TCP_FASTOPEN socket option (patch 6/8)
- TCP_FASTOPEN_KEY socket option (patch 7/8)
To support TFO for the server side, a few preparation patches are needed
(patches 2 to 5/8). Some of them were inspired by a previous work from
Benjamin Hesmans.
Note that TFO support with MPTCP has been validated with selftests
(patch 8/8) but also with Packetdrill tests running with a modified
but still very WIP version supporting MPTCP. Both the modified tool
and the tests are available online:
https://github.com/multipath-tcp/packetdrill/
====================
Link: https://lore.kernel.org/r/20221125222958.958636-1-matthieu.baerts@tessares.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This patch first adds TFO support in mptcp_connect.c.
This can be enabled via a new option: -o MPTFO.
Once enabled, the TCP_FASTOPEN socket option is enabled for the server
side and a sendto() with MSG_FASTOPEN is used instead of a connect() for
the client side.
Note that the first SYN has a limit of bytes it can carry. In other
words, it is allowed to send less data than the provided one. We then
need to track more status info to properly allow the next sendmsg()
starting from the next part of the data to send the rest.
Also in TFO scenarios, we need to completely spool the partially xmitted
buffer -- and account for that -- before starting sendfile/mmap xmit,
otherwise the relevant tests will fail.
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Dmytro Shytyi <dmytro@shytyi.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The goal of this socket option is to set different keys per listener,
see commit 1fba70e5b6 ("tcp: socket option to set TCP fast open key")
for more details about this socket option.
The only thing to do here with MPTCP is to relay the request to the
first subflow like it is already done for the other TCP_FASTOPEN* socket
options.
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The TCP_FASTOPEN socket option is one way for the application to tell
the kernel TFO support has to be enabled for the listener socket.
The only thing to do here with MPTCP is to relay the request to the
first subflow like it is already done for the other TCP_FASTOPEN* socket
options.
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Dmytro Shytyi <dmytro@shytyi.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The send_synack() needs to be overridden for MPTCP to support TFO for
two reasons:
- There is not be enough space in the TCP options if the TFO cookie has
to be added in the SYN+ACK with other options: MSS (4), SACK OK (2),
Timestamps (10), Window Scale (3+1), TFO (10+2), MP_CAPABLE (12).
MPTCPv1 specs -- RFC 8684, section B.1 [1] -- suggest to drop the TCP
timestamps option in this case.
- The data received in the SYN has to be handled: the SKB can be
dequeued from the subflow sk and transferred to the MPTCP sk. Counters
need to be updated accordingly and the application can be notified at
the end because some bytes have been received.
[1] https://www.rfc-editor.org/rfc/rfc8684.html#section-b.1
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Dmytro Shytyi <dmytro@shytyi.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently the initial ack sequence is generated on demand whenever
it's requested and the remote key is handy. The relevant code is
scattered in different places and can lead to multiple, unneeded,
crypto operations.
This change consolidates the ack sequence generation code in a single
helper, storing the sequence number at the subflow level.
The above additionally saves a few conditional in fast-path and will
simplify the upcoming fast-open implementation.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently in the receive path we don't need to discriminate
between MPC SYN, MPC SYN-ACK and MPC ACK, but soon the fastopen
code will need that info to properly track the fully established
status.
Track the exact MPC suboption type into the receive opt bitmap.
No functional change intended.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
64-bit RISC-V kernels have the kernel image mapped separately to alias
the linear map. The linear map and the kernel image map are documented
as "direct mapping" and "kernel" respectively in [1].
At image load time, the linear map corresponding to the kernel image
is set to PAGE_READ permission, and the kernel image map is set to
PAGE_READ|PAGE_EXEC.
When the initmem is freed, the pages in the linear map should be
restored to PAGE_READ|PAGE_WRITE, whereas the corresponding pages in
the kernel image map should be restored to PAGE_READ, by removing the
PAGE_EXEC permission.
This is not the case. For 64-bit kernels, only the linear map is
restored to its proper page permissions at initmem free, and not the
kernel image map.
In practise this results in that the kernel can potentially jump to
dead __init code, and start executing invalid instructions, without
getting an exception.
Restore the freed initmem properly, by setting both the kernel image
map to the correct permissions.
[1] Documentation/riscv/vm-layout.rst
Fixes: e5c35fa040 ("riscv: Map the kernel with correct permissions the first time")
Signed-off-by: Björn Töpel <bjorn@rivosinc.com>
Reviewed-by: Alexandre Ghiti <alex@ghiti.fr>
Tested-by: Alexandre Ghiti <alex@ghiti.fr>
Link: https://lore.kernel.org/r/20221115090641.258476-1-bjorn@kernel.org
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
lkp reported a build error, I tried the config and can reproduce
build error as below:
VDSOLD arch/riscv/kernel/vdso/vdso.so.dbg
ld.lld: error: section .note file range overlaps with .text
>>> .note range is [0x7C8, 0x803]
>>> .text range is [0x800, 0x1993]
ld.lld: error: section .text file range overlaps with .dynamic
>>> .text range is [0x800, 0x1993]
>>> .dynamic range is [0x808, 0x937]
ld.lld: error: section .note virtual address range overlaps with .text
>>> .note range is [0x7C8, 0x803]
>>> .text range is [0x800, 0x1993]
Fix it by setting DISABLE_BRANCH_PROFILING which will disable branch
tracing for vdso, thus avoid useless _ftrace_annotated_branch section
and _ftrace_branch section. Although we can also fix it by removing
the hardcoded .text begin address, but I think that's another story
and should be put into another patch.
Link: https://lore.kernel.org/lkml/202210122123.Cc4FPShJ-lkp@intel.com/#r
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Link: https://lore.kernel.org/r/20221102170254.1925-1-jszhang@kernel.org
Fixes: ad5d1122b8 ("riscv: use vDSO common flow to reduce the latency of the time-related functions")
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Currently, when detecting vmap stack overflow, riscv firstly switches
to the so called shadow stack, then use this shadow stack to call the
get_overflow_stack() to get the overflow stack. However, there's
a race here if two or more harts use the same shadow stack at the same
time.
To solve this race, we introduce spin_shadow_stack atomic var, which
will be swap between its own address and 0 in atomic way, when the
var is set, it means the shadow_stack is being used; when the var
is cleared, it means the shadow_stack isn't being used.
Fixes: 31da94c25a ("riscv: add VMAP_STACK overflow detection")
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Suggested-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Guo Ren <guoren@kernel.org>
Link: https://lore.kernel.org/r/20221030124517.2370-1-jszhang@kernel.org
[Palmer: Add AQ to the swap, and also some comments.]
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Pull networking fixes from Jakub Kicinski:
"Including fixes from bpf, can and wifi.
Current release - new code bugs:
- eth: mlx5e:
- use kvfree() in mlx5e_accel_fs_tcp_create()
- MACsec, fix RX data path 16 RX security channel limit
- MACsec, fix memory leak when MACsec device is deleted
- MACsec, fix update Rx secure channel active field
- MACsec, fix add Rx security association (SA) rule memory leak
Previous releases - regressions:
- wifi: cfg80211: don't allow multi-BSSID in S1G
- stmmac: set MAC's flow control register to reflect current settings
- eth: mlx5:
- E-switch, fix duplicate lag creation
- fix use-after-free when reverting termination table
Previous releases - always broken:
- ipv4: fix route deletion when nexthop info is not specified
- bpf: fix a local storage BPF map bug where the value's spin lock
field can get initialized incorrectly
- tipc: re-fetch skb cb after tipc_msg_validate
- wifi: wilc1000: fix Information Element parsing
- packet: do not set TP_STATUS_CSUM_VALID on CHECKSUM_COMPLETE
- sctp: fix memory leak in sctp_stream_outq_migrate()
- can: can327: fix potential skb leak when netdev is down
- can: add number of missing netdev freeing on error paths
- aquantia: do not purge addresses when setting the number of rings
- wwan: iosm:
- fix incorrect skb length leading to truncated packet
- fix crash in peek throughput test due to skb UAF"
* tag 'net-6.1-rc8-2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (79 commits)
net: ethernet: renesas: ravb: Fix promiscuous mode after system resumed
MAINTAINERS: Update maintainer list for chelsio drivers
ionic: update MAINTAINERS entry
sctp: fix memory leak in sctp_stream_outq_migrate()
packet: do not set TP_STATUS_CSUM_VALID on CHECKSUM_COMPLETE
net/mlx5: Lag, Fix for loop when checking lag
Revert "net/mlx5e: MACsec, remove replay window size limitation in offload path"
net: marvell: prestera: Fix a NULL vs IS_ERR() check in some functions
net: tun: Fix use-after-free in tun_detach()
net: mdiobus: fix unbalanced node reference count
net: hsr: Fix potential use-after-free
tipc: re-fetch skb cb after tipc_msg_validate
mptcp: fix sleep in atomic at close time
mptcp: don't orphan ssk in mptcp_close()
dsa: lan9303: Correct stat name
ipv4: Fix route deletion when nexthop info is not specified
net: wwan: iosm: fix incorrect skb length
net: wwan: iosm: fix crash in peek throughput test
net: wwan: iosm: fix dma_alloc_coherent incompatible pointer type
net: wwan: iosm: fix kernel test robot reported error
...
As the nla_nest_start() may fail with NULL returned, the return value
should be checked.
Note that this is not a real bug, nothing will break here.
The next nla_put() will fail as well and we'll bail (and
nla_nest_cancel() can handle NULL). But we keep getting
those "fixes" so whatever.
Signed-off-by: Yuan Can <yuancan@huawei.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://lore.kernel.org/r/20221129013934.55184-1-yuancan@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Per syzbot it is possible for perf_pending_task() to run after the
event is free()'d. There are two related but distinct cases:
- the task_work was already queued before destroying the event;
- destroying the event itself queues the task_work.
The first cannot be solved using task_work_cancel() since
perf_release() itself might be called from a task_work (____fput),
which means the current->task_works list is already empty and
task_work_cancel() won't be able to find the perf_pending_task()
entry.
The simplest alternative is extending the perf_event lifetime to cover
the task_work.
The second is just silly, queueing a task_work while you know the
event is going away makes no sense and is easily avoided by
re-arranging how the event is marked STATE_DEAD and ensuring it goes
through STATE_OFF on the way down.
Reported-by: syzbot+9228d6098455bb209ec8@syzkaller.appspotmail.com
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Marco Elver <elver@google.com>
This reverts commit c0071be0e1.
The cited commit removed the validity checks which initialized the
window_sz and never removed the use of the now uninitialized variable,
so now we are left with wrong value in the window size and the following
clang warning: [-Wuninitialized]
drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c:232:45:
warning: variable 'window_sz' is uninitialized when used here
MLX5_SET(macsec_aso, aso_ctx, window_size, window_sz);
Revet at this time to address the clang issue due to lack of time to
test the proper solution.
Fixes: c0071be0e1 ("net/mlx5e: MACsec, remove replay window size limitation in offload path")
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Reported-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20221129093006.378840-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
