Add new helper bpf_sk_storage_clone which optionally clones sk storage
and call it from sk_clone_lock.
Cc: Martin KaFai Lau <kafai@fb.com>
Cc: Yonghong Song <yhs@fb.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Rename existing bpf_map_inc_not_zero to __bpf_map_inc_not_zero to
indicate that it's caller's responsibility to do proper locking.
Create and export bpf_map_inc_not_zero wrapper that properly
locks map_idr_lock. Will be used in the next commit to
hold a map while cloning a socket.
Cc: Martin KaFai Lau <kafai@fb.com>
Cc: Yonghong Song <yhs@fb.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
There is a race in this test between receiving the ACK for the
single-byte packet sent in the test, and reading the values from the
map.
This patch fixes this by having the client wait until there are no more
unacknowledged packets.
Before:
for i in {1..1000}; do ../net/in_netns.sh ./test_tcp_rtt; \
done | grep -c PASSED
< trimmed error messages >
993
After:
for i in {1..10000}; do ../net/in_netns.sh ./test_tcp_rtt; \
done | grep -c PASSED
10000
Fixes: b55873984d ("selftests/bpf: test BPF_SOCK_OPS_RTT_CB")
Signed-off-by: Petar Penkov <ppenkov@google.com>
Reviewed-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
bpf_helpers.h and bpf_endian.h contain useful macros and BPF helper
definitions essential to almost every BPF program. Which makes them
useful not just for selftests. To be able to expose them as part of
libbpf, though, we need them to be dual-licensed as LGPL-2.1 OR
BSD-2-Clause. This patch updates licensing of those two files.
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Hechao Li <hechaol@fb.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Andrey Ignatov <rdna@fb.com>
Acked-by: Yonghong Song <yhs@fb.com>
Acked-by: Lawrence Brakmo <brakmo@fb.com>
Acked-by: Adam Barth <arb@fb.com>
Acked-by: Roman Gushchin <guro@fb.com>
Acked-by: Josef Bacik <jbacik@fb.com>
Acked-by: Joe Stringer <joe@wand.net.nz>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Acked-by: David Ahern <dsahern@gmail.com>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Ilya Leoshkevich <iii@linux.ibm.com>
Acked-by: Lorenz Bauer <lmb@cloudflare.com>
Acked-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Acked-by: Nikita V. Shirokov <tehnerd@tehnerd.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Acked-by: Petar Penkov <ppenkov@google.com>
Acked-by: Teng Qin <palmtenor@gmail.com>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Michal Rostecki <mrostecki@opensuse.org>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Sargun Dhillon <sargun@sargun.me>
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Don't uninstall an XDP program when none is installed, and don't install
an XDP program that has the same ID as the one already installed.
dev_change_xdp_fd doesn't perform any checks in case it uninstalls an
XDP program. It means that the driver's ndo_bpf can be called with
XDP_SETUP_PROG asking to set it to NULL even if it's already NULL. This
case happens if the user runs `ip link set eth0 xdp off` when there is
no XDP program attached.
The symmetrical case is possible when the user tries to set the program
that is already set.
The drivers typically perform some heavy operations on XDP_SETUP_PROG,
so they all have to handle these cases internally to return early if
they happen. This patch puts this check into the kernel code, so that
all drivers will benefit from it.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Magnus Karlsson says:
====================
This patch set adds support for a new flag called need_wakeup in the
AF_XDP Tx and fill rings. When this flag is set by the driver, it
means that the application has to explicitly wake up the kernel Rx
(for the bit in the fill ring) or kernel Tx (for bit in the Tx ring)
processing by issuing a syscall. Poll() can wake up both and sendto()
will wake up Tx processing only.
The main reason for introducing this new flag is to be able to
efficiently support the case when application and driver is executing
on the same core. Previously, the driver was just busy-spinning on the
fill ring if it ran out of buffers in the HW and there were none to
get from the fill ring. This approach works when the application and
driver is running on different cores as the application can replenish
the fill ring while the driver is busy-spinning. Though, this is a
lousy approach if both of them are running on the same core as the
probability of the fill ring getting more entries when the driver is
busy-spinning is zero. With this new feature the driver now sets the
need_wakeup flag and returns to the application. The application can
then replenish the fill queue and then explicitly wake up the Rx
processing in the kernel using the syscall poll(). For Tx, the flag is
only set to one if the driver has no outstanding Tx completion
interrupts. If it has some, the flag is zero as it will be woken up by
a completion interrupt anyway. This flag can also be used in other
situations where the driver needs to be woken up explicitly.
As a nice side effect, this new flag also improves the Tx performance
of the case where application and driver are running on two different
cores as it reduces the number of syscalls to the kernel. The kernel
tells user space if it needs to be woken up by a syscall, and this
eliminates many of the syscalls. The Rx performance of the 2-core case
is on the other hand slightly worse, since there is a need to use a
syscall now to wake up the driver, instead of the driver
busy-spinning. It does waste less CPU cycles though, which might lead
to better overall system performance.
This new flag needs some simple driver support. If the driver does not
support it, the Rx flag is always zero and the Tx flag is always
one. This makes any application relying on this feature default to the
old behavior of not requiring any syscalls in the Rx path and always
having to call sendto() in the Tx path.
For backwards compatibility reasons, this feature has to be explicitly
turned on using a new bind flag (XDP_USE_NEED_WAKEUP). I recommend
that you always turn it on as it has a large positive performance
impact for the one core case and does not degrade 2 core performance
and actually improves it for Tx heavy workloads.
Here are some performance numbers measured on my local,
non-performance optimized development system. That is why you are
seeing numbers lower than the ones from Björn and Jesper. 64 byte
packets at 40Gbit/s line rate. All results in Mpps. Cores == 1 means
that both application and driver is executing on the same core. Cores
== 2 that they are on different cores.
Applications
need_wakeup cores txpush rxdrop l2fwd
---------------------------------------------------------------
n 1 0.07 0.06 0.03
y 1 21.6 8.2 6.5
n 2 32.3 11.7 8.7
y 2 33.1 11.7 8.7
Overall, the need_wakeup flag provides the same or better performance
in all the micro-benchmarks. The reduction of sendto() calls in txpush
is large. Only a few per second is needed. For l2fwd, the drop is 50%
for the 1 core case and more than 99.9% for the 2 core case. Do not
know why I am not seeing the same drop for the 1 core case yet.
The name and inspiration of the flag has been taken from io_uring by
Jens Axboe. Details about this feature in io_uring can be found in
http://kernel.dk/io_uring.pdf, section 8.3. It also addresses most of
the denial of service and sendto() concerns raised by Maxim
Mikityanskiy in https://www.spinics.net/lists/netdev/msg554657.html.
The typical Tx part of an application will have to change from:
ret = sendto(fd,....)
to:
if (xsk_ring_prod__needs_wakeup(&xsk->tx))
ret = sendto(fd,....)
and th Rx part from:
rcvd = xsk_ring_cons__peek(&xsk->rx, BATCH_SIZE, &idx_rx);
if (!rcvd)
return;
to:
rcvd = xsk_ring_cons__peek(&xsk->rx, BATCH_SIZE, &idx_rx);
if (!rcvd) {
if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
ret = poll(fd,.....);
return;
}
v3 -> v4:
* Maxim found a possible race in the Tx part of the driver. The
setting of the flag needs to happen before the sending, otherwise it
might trigger this race. Fixed in ixgbe and i40e driver.
* Mellanox support contributed by Maxim
* Removed the XSK_DRV_CAN_SLEEP flag as it was not used
anymore. Thanks to Sridhar for discovering this.
* For consistency the feature is now always called need_wakeup. There
were some places where it was referred to as might_sleep, but they
have been removed. Thanks to Sridhar for spotting.
* Fixed some typos in the commit messages
v2 -> v3:
* Converted the Mellanox driver to the new ndo in patch 1 as pointed
out by Maxim
* Fixed the compatibility code of XDP_MMAP_OFFSETS so it now works.
v1 -> v2:
* Fixed bisectability problem pointed out by Jakub
* Added missing initiliztion of the Tx need_wakeup flag to 1
This patch has been applied against commit b753c5a7f9 ("Merge branch 'r8152-RX-improve'")
Structure of the patch set:
Patch 1: Replaces the ndo_xsk_async_xmit with ndo_xsk_wakeup to
support waking up both Rx and Tx processing
Patch 2: Implements the need_wakeup functionality in common code
Patch 3-4: Add need_wakeup support to the i40e and ixgbe drivers
Patch 5: Add need_wakeup support to libbpf
Patch 6: Add need_wakeup support to the xdpsock sample application
Patch 7-8: Add need_wakeup support to the Mellanox mlx5 driver
====================
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This commit adds support for the new need_wakeup feature of AF_XDP. The
applications can opt-in by using the XDP_USE_NEED_WAKEUP bind() flag.
When this feature is enabled, some behavior changes:
RX side: If the Fill Ring is empty, instead of busy-polling, set the
flag to tell the application to kick the driver when it refills the Fill
Ring.
TX side: If there are pending completions or packets queued for
transmission, set the flag to tell the application that it can skip the
sendto() syscall and save time.
The performance testing was performed on a machine with the following
configuration:
- 24 cores of Intel Xeon E5-2620 v3 @ 2.40 GHz
- Mellanox ConnectX-5 Ex with 100 Gbit/s link
The results with retpoline disabled:
| without need_wakeup | with need_wakeup |
|----------------------|----------------------|
| one core | two cores | one core | two cores |
-------|----------|-----------|----------|-----------|
txonly | 20.1 | 33.5 | 29.0 | 34.2 |
rxdrop | 0.065 | 14.1 | 12.0 | 14.1 |
l2fwd | 0.032 | 7.3 | 6.6 | 7.2 |
"One core" means the application and NAPI run on the same core. "Two
cores" means they are pinned to different cores.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Saeed Mahameed <saeedm@mellanox.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Two XSK tasks are performed during NAPI polling, that are not bound to
hardware interrupts: TXing packets and polling for frames in the Fill
Ring. They are special in a way that the hardware doesn't know about
these tasks, so it doesn't trigger interrupts if there is still some
work to be done, it's our driver's responsibility to ensure NAPI will be
rescheduled if needed.
Create a new function to handle these tasks and move the corresponding
code from mlx5e_napi_poll to the new function to improve modularity and
prepare for the changes in the following patch.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Saeed Mahameed <saeedm@mellanox.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This commit adds using the need_wakeup flag to the xdpsock sample
application. It is turned on by default as we think it is a feature
that seems to always produce a performance benefit, if the application
has been written taking advantage of it. It can be turned off in the
sample app by using the '-m' command line option.
The txpush and l2fwd sub applications have also been updated to
support poll() with multiple sockets.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This commit adds support for the new need_wakeup flag in AF_XDP. The
xsk_socket__create function is updated to handle this and a new
function is introduced called xsk_ring_prod__needs_wakeup(). This
function can be used by the application to check if Rx and/or Tx
processing needs to be explicitly woken up.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This patch adds support for the need_wakeup feature of AF_XDP. If the
application has told the kernel that it might sleep using the new bind
flag XDP_USE_NEED_WAKEUP, the driver will then set this flag if it has
no more buffers on the NIC Rx ring and yield to the application. For
Tx, it will set the flag if it has no outstanding Tx completion
interrupts and return to the application.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This patch adds support for the need_wakeup feature of AF_XDP. If the
application has told the kernel that it might sleep using the new bind
flag XDP_USE_NEED_WAKEUP, the driver will then set this flag if it has
no more buffers on the NIC Rx ring and yield to the application. For
Tx, it will set the flag if it has no outstanding Tx completion
interrupts and return to the application.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This commit adds support for a new flag called need_wakeup in the
AF_XDP Tx and fill rings. When this flag is set, it means that the
application has to explicitly wake up the kernel Rx (for the bit in
the fill ring) or kernel Tx (for bit in the Tx ring) processing by
issuing a syscall. Poll() can wake up both depending on the flags
submitted and sendto() will wake up tx processing only.
The main reason for introducing this new flag is to be able to
efficiently support the case when application and driver is executing
on the same core. Previously, the driver was just busy-spinning on the
fill ring if it ran out of buffers in the HW and there were none on
the fill ring. This approach works when the application is running on
another core as it can replenish the fill ring while the driver is
busy-spinning. Though, this is a lousy approach if both of them are
running on the same core as the probability of the fill ring getting
more entries when the driver is busy-spinning is zero. With this new
feature the driver now sets the need_wakeup flag and returns to the
application. The application can then replenish the fill queue and
then explicitly wake up the Rx processing in the kernel using the
syscall poll(). For Tx, the flag is only set to one if the driver has
no outstanding Tx completion interrupts. If it has some, the flag is
zero as it will be woken up by a completion interrupt anyway.
As a nice side effect, this new flag also improves the performance of
the case where application and driver are running on two different
cores as it reduces the number of syscalls to the kernel. The kernel
tells user space if it needs to be woken up by a syscall, and this
eliminates many of the syscalls.
This flag needs some simple driver support. If the driver does not
support this, the Rx flag is always zero and the Tx flag is always
one. This makes any application relying on this feature default to the
old behaviour of not requiring any syscalls in the Rx path and always
having to call sendto() in the Tx path.
For backwards compatibility reasons, this feature has to be explicitly
turned on using a new bind flag (XDP_USE_NEED_WAKEUP). I recommend
that you always turn it on as it so far always have had a positive
performance impact.
The name and inspiration of the flag has been taken from io_uring by
Jens Axboe. Details about this feature in io_uring can be found in
http://kernel.dk/io_uring.pdf, section 8.3.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
This commit replaces ndo_xsk_async_xmit with ndo_xsk_wakeup. This new
ndo provides the same functionality as before but with the addition of
a new flags field that is used to specifiy if Rx, Tx or both should be
woken up. The previous ndo only woke up Tx, as implied by the
name. The i40e and ixgbe drivers (which are all the supported ones)
are updated with this new interface.
This new ndo will be used by the new need_wakeup functionality of XDP
sockets that need to be able to wake up both Rx and Tx driver
processing.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Jose Abreu says:
====================
net: stmmac: Improvements for -next
Couple of improvements for -next tree. More info in commit logs.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Add 2 new selftests for VLAN Insertion offloading. Tests are for inner
and outer VLAN offloading.
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Adds the logic to insert a given VLAN ID in a packet. This is offloaded
to HW and its descriptor based. For now, only XGMAC implements the
necessary callbacks.
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for EEE in XGMAC cores by implementing the necessary
callbacks.
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add 4 new tests:
- SA Insertion (register based)
- SA Insertion (descriptor based)
- SA Replacament (register based)
- SA Replacement (descriptor based)
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add the support for Source Address Insertion and Replacement in XGMAC
cores. Two methods are supported: Descriptor based and register based.
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add the ethtool interface to dump the register map in XGMAC cores.
Changes from v2:
- Remove uneeded memset (Jakub)
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a counter that increments each time a packet with split header is
received.
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add the support for Split Header feature in the RX path and enable it in
XGMAC cores.
This does not impact neither beneficts bandwidth but it does reduces CPU
usage because without the feature all the entire packet is memcpy'ed,
while that with the feature only the header is.
With Split Header disabled 'perf stat -d' gives:
86870.624945 task-clock (msec) # 0.429 CPUs utilized
1073352 context-switches # 0.012 M/sec
1 cpu-migrations # 0.000 K/sec
213 page-faults # 0.002 K/sec
327113872376 cycles # 3.766 GHz (62.53%)
56618161216 instructions # 0.17 insn per cycle (75.06%)
10742205071 branches # 123.658 M/sec (75.36%)
584309242 branch-misses # 5.44% of all branches (75.19%)
17594787965 L1-dcache-loads # 202.540 M/sec (74.88%)
4003773131 L1-dcache-load-misses # 22.76% of all L1-dcache hits (74.89%)
1313301468 LLC-loads # 15.118 M/sec (49.75%)
355906510 LLC-load-misses # 27.10% of all LL-cache hits (49.92%)
With Split Header enabled 'perf stat -d' gives:
49324.456539 task-clock (msec) # 0.245 CPUs utilized
2542387 context-switches # 0.052 M/sec
1 cpu-migrations # 0.000 K/sec
213 page-faults # 0.004 K/sec
177092791469 cycles # 3.590 GHz (62.30%)
68555756017 instructions # 0.39 insn per cycle (75.16%)
12697019382 branches # 257.418 M/sec (74.81%)
442081897 branch-misses # 3.48% of all branches (74.79%)
20337958358 L1-dcache-loads # 412.330 M/sec (75.46%)
3820210140 L1-dcache-load-misses # 18.78% of all L1-dcache hits (75.35%)
1257719198 LLC-loads # 25.499 M/sec (49.73%)
685543923 LLC-load-misses # 54.51% of all LL-cache hits (49.86%)
Changes from v2:
- Reword commit message (Jakub)
Changes from v1:
- Add performance info (David)
- Add misssing dma_sync_single_for_device()
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Return the correct value when RX descriptor is not the last one.
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In order to add Split Header support, stmmac_rx() needs to take into
account that packet may be split accross multiple descriptors.
Refactor the logic of this function in order to support this scenario.
Changes from v2:
- Fixup if condition detection (Jakub)
- Don't stop NAPI with unfinished packet (Jakub)
- Use napi_alloc_skb() (Jakub)
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TX Timestamp in XGMAC comes from MAC instead of descriptors. Implement
this in a new callback.
Also, RX Timestamp in XGMAC must be cheked against corruption and we need
a barrier to make sure that descriptor fields are read correctly.
Changes from v2:
- Rework return code check (Jakub)
Changes from v1:
- Rework the get timestamp function (David)
Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ido Schimmel says:
====================
Add drop monitor for offloaded data paths
Users have several ways to debug the kernel and understand why a packet
was dropped. For example, using drop monitor and perf. Both utilities
trace kfree_skb(), which is the function called when a packet is freed
as part of a failure. The information provided by these tools is
invaluable when trying to understand the cause of a packet loss.
In recent years, large portions of the kernel data path were offloaded
to capable devices. Today, it is possible to perform L2 and L3
forwarding in hardware, as well as tunneling (IP-in-IP and VXLAN).
Different TC classifiers and actions are also offloaded to capable
devices, at both ingress and egress.
However, when the data path is offloaded it is not possible to achieve
the same level of introspection since packets are dropped by the
underlying device and never reach the kernel.
This patchset aims to solve this by allowing users to monitor packets
that the underlying device decided to drop along with relevant metadata
such as the drop reason and ingress port.
The above is achieved by exposing a fundamental capability of devices
capable of data path offloading - packet trapping. In much the same way
as drop monitor registers its probe function with the kfree_skb()
tracepoint, the device is instructed to pass to the CPU (trap) packets
that it decided to drop in various places in the pipeline.
The configuration of the device to pass such packets to the CPU is
performed using devlink, as it is not specific to a port, but rather to
a device. In the future, we plan to control the policing of such packets
using devlink, in order not to overwhelm the CPU.
While devlink is used as the control path, the dropped packets are
passed along with metadata to drop monitor, which reports them to
userspace as netlink events. This allows users to use the same interface
for the monitoring of both software and hardware drops.
Logically, the solution looks as follows:
Netlink event: Packet w/ metadata
Or a summary of recent drops
^
|
Userspace |
+---------------------------------------------------+
Kernel |
|
+-------+--------+
| |
| drop_monitor |
| |
+-------^--------+
|
|
|
+----+----+
| | Kernel's Rx path
| devlink | (non-drop traps)
| |
+----^----+ ^
| |
+-----------+
|
+-------+-------+
| |
| Device driver |
| |
+-------^-------+
Kernel |
+---------------------------------------------------+
Hardware |
| Trapped packet
|
+--+---+
| |
| ASIC |
| |
+------+
In order to reduce the patch count, this patchset only includes
integration with netdevsim. A follow-up patchset will add devlink-trap
support in mlxsw.
Patches #1-#7 extend drop monitor to also monitor hardware originated
drops.
Patches #8-#10 add the devlink-trap infrastructure.
Patches #11-#12 add devlink-trap support in netdevsim.
Patches #13-#16 add tests for the generic infrastructure over netdevsim.
Example
=======
Instantiate netdevsim
---------------------
List supported traps
--------------------
netdevsim/netdevsim10:
name source_mac_is_multicast type drop generic true action drop group l2_drops
name vlan_tag_mismatch type drop generic true action drop group l2_drops
name ingress_vlan_filter type drop generic true action drop group l2_drops
name ingress_spanning_tree_filter type drop generic true action drop group l2_drops
name port_list_is_empty type drop generic true action drop group l2_drops
name port_loopback_filter type drop generic true action drop group l2_drops
name fid_miss type exception generic false action trap group l2_drops
name blackhole_route type drop generic true action drop group l3_drops
name ttl_value_is_too_small type exception generic true action trap group l3_drops
name tail_drop type drop generic true action drop group buffer_drops
Enable a trap
-------------
Query statistics
----------------
netdevsim/netdevsim10:
name blackhole_route type drop generic true action trap group l3_drops
stats:
rx:
bytes 7384 packets 52
Monitor dropped packets
-----------------------
dropwatch> set alertmode packet
Setting alert mode
Alert mode successfully set
dropwatch> set sw true
setting software drops monitoring to 1
dropwatch> set hw true
setting hardware drops monitoring to 1
dropwatch> start
Enabling monitoring...
Kernel monitoring activated.
Issue Ctrl-C to stop monitoring
drop at: ttl_value_is_too_small (l3_drops)
origin: hardware
input port ifindex: 55
input port name: eth0
timestamp: Mon Aug 12 10:52:20 2019 445911505 nsec
protocol: 0x800
length: 142
original length: 142
drop at: ip6_mc_input+0x8b8/0xef8 (0xffffffff9e2bb0e8)
origin: software
input port ifindex: 4
timestamp: Mon Aug 12 10:53:37 2019 024444587 nsec
protocol: 0x86dd
length: 110
original length: 110
Future plans
============
* Provide more drop reasons as well as more metadata
* Add dropmon support to libpcap, so that tcpdump/tshark could
specifically listen on dropmon traffic, instead of capturing all
netlink packets via nlmon interface
Changes in v3:
* Place test with the rest of the netdevsim tests
* Fix test to load netdevsim module
* Move devlink helpers from the test to devlink_lib.sh. Will be used
by mlxsw tests
* Re-order netdevsim includes in alphabetical order
* Fix reverse xmas tree in netdevsim
* Remove double include in netdevsim
Changes in v2:
* Use drop monitor to report dropped packets instead of devlink
* Add drop monitor patches
* Add test cases
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Add test cases for devlink-trap on top of the netdevsim implementation.
The tests focus on the devlink-trap core infrastructure and user space
API. They test both good and bad flows and also dismantle of the netdev
and devlink device used to report trapped packets.
This allows device drivers to focus their tests on device-specific
functionality.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add helpers to interact with devlink-trap, such as setting the action of
a trap and retrieving statistics.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For tests that create their network interfaces dynamically or do not use
interfaces at all (as with netdevsim) it is useful to define their own
devlink device instead of deriving it from the first network interface.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Have netdevsim register its trap groups and traps with devlink during
initialization and periodically report trapped packets to devlink core.
Since netdevsim is not a real device, the trapped packets are emulated
using a workqueue that periodically reports a UDP packet with a random
5-tuple from each active packet trap and from each running netdev.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add initial documentation of the devlink-trap mechanism, explaining the
background, motivation and the semantics of the interface.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add generic packet traps and groups that can report dropped packets as
well as exceptions such as TTL error.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add the basic packet trap infrastructure that allows device drivers to
register their supported packet traps and trap groups with devlink.
Each driver is expected to provide basic information about each
supported trap, such as name and ID, but also the supported metadata
types that will accompany each packet trapped via the trap. The
currently supported metadata type is just the input port, but more will
be added in the future. For example, output port and traffic class.
Trap groups allow users to set the action of all member traps. In
addition, users can retrieve per-group statistics in case per-trap
statistics are too narrow. In the future, the trap group object can be
extended with more attributes, such as policer settings which will limit
the amount of traffic generated by member traps towards the CPU.
Beside registering their packet traps with devlink, drivers are also
expected to report trapped packets to devlink along with relevant
metadata. devlink will maintain packets and bytes statistics for each
packet trap and will potentially report the trapped packet with its
metadata to user space via drop monitor netlink channel.
The interface towards the drivers is simple and allows devlink to set
the action of the trap. Currently, only two actions are supported:
'trap' and 'drop'. When set to 'trap', the device is expected to provide
the sole copy of the packet to the driver which will pass it to devlink.
When set to 'drop', the device is expected to drop the packet and not
send a copy to the driver. In the future, more actions can be added,
such as 'mirror'.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Drop monitor has start and stop commands, but so far these were only
used to start and stop monitoring of software drops.
Now that drop monitor can also monitor hardware drops, we should allow
the user to control these as well.
Do that by adding SW and HW flags to these commands. If no flag is
specified, then only start / stop monitoring software drops. This is
done in order to maintain backward-compatibility with existing user
space applications.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In summary alert mode a notification is sent with a list of recent drop
reasons and a count of how many packets were dropped due to this reason.
To avoid expensive operations in the context in which packets are
dropped, each CPU holds an array whose number of entries is the maximum
number of drop reasons that can be encoded in the netlink notification.
Each entry stores the drop reason and a count. When a packet is dropped
the array is traversed and a new entry is created or the count of an
existing entry is incremented.
Later, in process context, the array is replaced with a newly allocated
copy and the old array is encoded in a netlink notification. To avoid
breaking user space, the notification includes the ancillary header,
which is 'struct net_dm_alert_msg' with number of entries set to '0'.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In a similar fashion to software drops, extend drop monitor to send
netlink events when packets are dropped by the underlying hardware.
The main difference is that instead of encoding the program counter (PC)
from which kfree_skb() was called in the netlink message, we encode the
hardware trap name. The two are mostly equivalent since they should both
help the user understand why the packet was dropped.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The drop monitor configuration (e.g., alert mode) is global, but user
will be able to enable monitoring of only software or hardware drops.
Therefore, ensure that monitoring of both software and hardware drops are
disabled before allowing drop monitor configuration to take place.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Export a function that can be invoked in order to report packets that
were dropped by the underlying hardware along with metadata.
Subsequent patches will add support for the different alert modes.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Like software drops, hardware drops also need the same type of per-CPU
data. Therefore, initialize it during module initialization and
de-initialize it during module exit.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently drop monitor only reports software drops to user space, but
subsequent patches are going to add support for hardware drops.
Like software drops, the per-CPU data of hardware drops needs to be
initialized and de-initialized upon module initialization and exit. To
avoid code duplication, break this code into separate functions, so that
these could be re-used for hardware drops.
No functional changes intended.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Johan Hedberg says:
====================
pull request: bluetooth 2019-08-17
Here's a set of Bluetooth fixes for the 5.3-rc series:
- Multiple fixes for Qualcomm (btqca & hci_qca) drivers
- Minimum encryption key size debugfs setting (this is required for
Bluetooth Qualification)
- Fix hidp_send_message() to have a meaningful return value
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Nikolay Aleksandrov says:
====================
net: bridge: mdb: allow dump/add/del of host-joined entries
This set makes the bridge dump host-joined mdb entries, they should be
treated as normal entries since they take a slot and are aging out.
We already have notifications for them but we couldn't dump them until
now so they remained hidden. We dump them similar to how they're
notified, in order to keep user-space compatibility with the dumped
objects (e.g. iproute2 dumps mdbs in a format which can be fed into
add/del commands) we allow host-joined groups also to be added/deleted via
mdb commands. That can later be used for L2 mcast MAC manipulation as
was recently discussed. Note that iproute2 changes are not necessary,
this set will work with the current user-space mdb code.
Patch 01 - a trivial comment move
Patch 02 - factors out the mdb filling code so it can be
re-used for the host-joined entries
Patch 03 - dumps host-joined entries
Patch 04 - allows manipulation of host-joined entries via standard mdb
calls
v3: fix compiler warning in patch 04 (DaveM)
v2: change patch 04 to avoid double notification and improve host group
manual removal if no ports are present in the group
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently this is needed only for user-space compatibility, so similar
object adds/deletes as the dumped ones would succeed. Later it can be
used for L2 mcast MAC add/delete.
v3: fix compiler warning (DaveM)
v2: don't send a notification when used from user-space, arm the group
timer if no ports are left after host entry del
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently we dump only the port mdb entries but we can have host-joined
entries on the bridge itself and they should be treated as normal temp
mdbs, they're already notified:
$ bridge monitor all
[MDB]dev br0 port br0 grp ff02::8 temp
The group will not be shown in the bridge mdb output, but it takes 1 slot
and it's timing out. If it's only host-joined then the mdb show output
can even be empty.
After this patch we show the host-joined groups:
$ bridge mdb show
dev br0 port br0 grp ff02::8 temp
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We have to factor out the mdb fill portion in order to re-use it later for
the bridge mdb entries. No functional changes intended.
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Trivial patch to move the vlan comments in their proper places above the
vid 0 checks.
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
