Every matcher has RX and TX paths. When a new matcher is created, its RX
and TX start/end anchors are connected to the respective RX and TX anchors
of the previous and next matchers.
This creates a potential performance issue: when a certain rule is added
to a matcher, in many cases it is RX or TX only rule, which may create a
long chain of RX/TX-only paths w/o the actual rules.
This patch aims to handle this issue.
RX and TX matchers are now handled separately: matcher connection in the
matchers chain is split into two separate lists: RX only and TX only.
when a new matcher is created, it is initially created 'detached' - its
RX/TX members are not inserted into the table's matcher list.
When an actual rule is added, only its appropriate RX or TX nic matchers
are then added to the table's nic matchers list and inserted into its
place in the chain of matchers.
I.e., if the rule that is being added is an RX-only rule, only the RX
part of the matcher will be connected to the chain, while TX part of the
matcher remains detached and doesn't prolong the TX chain of the matchers.
Same goes for rule deletion: when the last RX/TX rule of the nic matcher
is destroyed, the nic matcher is removed from its list.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Match on geneve_tlv_option_0_exist field on devices that support STEv1.
Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Tunnel headers are generic encapsulation headers, applies for all
tunneling protocols identified by the device native parser or by the
programmable parser, this support will enable raw matching headers 0 and 1.
Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Add support for misc5 match parameter as per HW spec, this will allow
matching on tunnel_header fields.
Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Extend mlx5 debugfs support to present Software Steering resources:
dr_domain including it's tables, matchers and rules.
The interface is read-only. While dump is being presented, new steering
rules cannot be inserted/deleted.
The steering information is dumped in the CSV form with the following
format:
<object_type>,<object_ID>, <object_info>,...,<object_info>
This data can be read at the following path:
/sys/kernel/debug/mlx5/<BDF>/steering/fdb/<domain_handle>
Example:
# cat /sys/kernel/debug/mlx5/0000:82:00.0/steering/fdb/dmn_000018644
3100,0x55caa4621c50,0xee802,4,65533
3101,0x55caa4621c50,0xe0100008
Changes in V2:
- Reduce temp hex buffer size and avoid unnecessary memset
- Use bin2hex() instead of DIY loop
- Don't check debugfs functions return values
Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Allow only legal values for flex parser ID - values from 0 to 7.
For other values skip the parser, and as a result the matcher creation
will fail for using invalid flex parser ID.
Signed-off-by: Hamdan Igbaria <hamdani@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
In dr_types structs, some list fields are list heads, and some
are just list nodes that are stored on the other structs' lists.
Rename the appropriate list field to reflect this distinction.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Macros prefix should be capital letters - fix the prefix in
mlx5_FLEX_PARSER_MPLS_OVER_UDP_ENABLED.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
The error code of nic matcher init functions wasn't checked.
This patch improves the matcher init function and fix error flow bug:
the handling of match parameter is moved into a separate function
and error flow is simplified.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
commit 077cdda764 ("net/mlx5e: TC, Fix memory leak with rules with internal port")
commit 31108d142f ("net/mlx5: Fix some error handling paths in 'mlx5e_tc_add_fdb_flow()'")
commit 4390c6edc0 ("net/mlx5: Fix some error handling paths in 'mlx5e_tc_add_fdb_flow()'")
https://lore.kernel.org/all/20211229065352.30178-1-saeed@kernel.org/
net/smc/smc_wr.c
commit 49dc9013e3 ("net/smc: Use the bitmap API when applicable")
commit 349d43127d ("net/smc: fix kernel panic caused by race of smc_sock")
bitmap_zero()/memset() is removed by the fix
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The reference taken by 'of_find_device_by_node()' must be released when
not needed anymore.
Add the corresponding 'put_device()' in the and error handling paths.
Fixes: 18a6c85fcc ("fsl/fman: Add FMan Port Support")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As reported by checkpatch.pl, no space is necessary after a cast.
Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
checkpatch.pl complains as the following:
Alignment should match open parenthesis
Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
Documentation/process/coding-style.rst says (in line 88)
"Don't put multiple assignments on a single line either."
This patch fixes the coding style issue reported by checkpatch.pl.
Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add () around macro argument to avoid precedence issues
Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
strlcpy is marked as deprecated in Documentation/process/deprecated.rst,
and there is no functional difference when the caller expects truncation
(when not checking the return value). strscpy is relatively better as it
also avoids scanning the whole source string.
This silences the related checkpatch warnings from:
commit 5dbdb2d87c ("checkpatch: prefer strscpy to strlcpy")
Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support to enable flow-director filter when multiple TCs are
configured. Flow director filter can be configured using ethtool
(--config-ntuple option). When multiple TCs are configured, each
TC is mapped to an unique HW VSI. So VSI corresponding to queue
used in filter is identified and flow director context is updated
with correct VSI while configuring ntuple filter in HW.
Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
queue
Tony Nguyen says:
====================
1GbE Intel Wired LAN Driver Updates 2021-12-29
Ruud Bos says:
The igb driver provides support for PEROUT and EXTTS pin functions that
allow adapter external use of timing signals. At Hottinger Bruel & Kjaer we
are using the PEROUT function to feed a PTP corrected 1pps signal into an
FPGA as cross system synchronized time source.
Support for the PEROUT and EXTTS SDP functions is currently limited to
i210/i211 based adapters. This patch series enables these functions also
for 82580/i354/i350 based ones. Because the time registers of these
adapters do not have the nice split in second rollovers as the i210 has,
the implementation is slightly more complex compared to the i210
implementation.
The PEROUT function has been successfully tested on an i350 based ethernet
adapter. Using the following user space code excerpt, the driver outputs a
PTP corrected 1pps signal on the SDP0 pin of an i350:
struct ptp_pin_desc desc;
memset(&desc, 0, sizeof(desc));
desc.index = 0;
desc.func = PTP_PF_PEROUT;
desc.chan = 0;
if (ioctl(fd, PTP_PIN_SETFUNC, &desc) == 0) {
struct timespec ts;
if (clock_gettime(clkid, &ts) == 0) {
struct ptp_perout_request rq;
memset(&rq, 0, sizeof(rq));
rq.index = 0;
rq.start.sec = ts.tv_sec + 1;
rq.start.nsec = 500000000;
rq.period.sec = 1;
rq.period.nsec = 0;
if (ioctl(fd, PTP_PEROUT_REQUEST, &rq) == 0) {
/* 1pps signal is now available on SDP0 */
}
}
}
The added EXTTS function has not been tested. However, looking at the data
sheets, the layout of the registers involved match the i210 exactly except
for the time registers mentioned before. Hence the almost identical
implementation.
---
Note: I made changes to fix RCT and checkpatch messages regarding
unnecessary parenthesis.
====================
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add inetaddr notifiers to support add/del IPv4 address on switchdev
port. We create TRAP on first address, added on port and delete TRAP,
when last address removed.
Currently, driver supports only regular port to became routed.
Other port type support will be added later
Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Initial implementation of notification handlers. For now this is just
stub.
So that we can move forward and add prestera_router_hw's objects
manipulations.
We support several addresses on interface. We just have nothing to do for
second address, because rif is already enabled on this interface, after
first one.
Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add prestera_router_hw.c. This file contains functions, which track HW
objects relations and links. This include implicity creation of objects,
that needed by requested one and implicity removing of objects, which
reference counter is became zero.
We need this layer, because kernel callbacks not always mapped to
creation of single HW object. So let it be two different layers - one
for subscribing and parsing kernel structures, and another
(prestera_router_hw.c) for HW objects relations tracking.
There is two types of objects on router_hw layer:
- Explicit objects (rif_entry) : created by higher layer.
- Implicit objects (vr) : created on demand by explicit objects.
Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add prestera_router.c, which contains code to subscribe/unsubscribe on
kernel notifiers for router. This handle kernel notifications,
parse structures to make key to manipulate prestera_router_hw's objects.
Also prestera_router is container for router's objects database.
Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add functions and structures to allocate virtual router.
prestera_hw_vr_create() return index of allocated VR so that we can move
forward and also add another objects (e.g. router interface),
which has link to VR.
Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Thanks for your review. Here is the new version for this patch.
This patch adds support for the emac rx dma present on sun4i. The emac
is able to move packets from rx fifo to RAM by using dma.
Change since v4.
- rename sbk field to skb
- rename alloc_emac_dma_req to emac_alloc_dma_req
- using kzalloc(..., GPF_ATOMIC) in interrupt context to avoid
sleeping
- retry by using emac_inblk_32bit when emac_dma_inblk_32bit fails
- fix some code style issues
Change since v5.
- fix some code style issue
Signed-off-by: Conley Lee <conleylee@foxmail.com>
Link: https://lore.kernel.org/r/tencent_DE05ADA53D5B084D4605BE6CB11E49EF7408@qq.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Support for the PTP pin function on 82580/i354/i350 based adapters.
Because the time registers of these adapters do not have the nice split in
second rollovers as the i210 has, the implementation is slightly more
complex compared to the i210 implementation.
Signed-off-by: Ruud Bos <kernel.hbk@gmail.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Support for the PEROUT PTP pin function on 82580/i354/i350 based adapters.
Because the time registers of these adapters do not have the nice split in
second rollovers as the i210 has, the implementation is slightly more
complex compared to the i210 implementation.
Signed-off-by: Ruud Bos <kernel.hbk@gmail.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Remove code duplication in the tsync interrupt handler function by moving
this logic to separate functions. This keeps the interrupt handler readable
and allows the new functions to be extended for adapter types other than
i210.
Signed-off-by: Ruud Bos <kernel.hbk@gmail.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
sock.h is pretty heavily used (5k objects rebuilt on x86 after
it's touched). We can drop the include of filter.h from it and
add a forward declaration of struct sk_filter instead.
This decreases the number of rebuilt objects when bpf.h
is touched from ~5k to ~1k.
There's a lot of missing includes this was masking. Primarily
in networking tho, this time.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Nikolay Aleksandrov <nikolay@nvidia.com>
Acked-by: Stefano Garzarella <sgarzare@redhat.com>
Link: https://lore.kernel.org/bpf/20211229004913.513372-1-kuba@kernel.org
In case of an error in mlx5e_set_features(), 'netdev->features' must be
updated with the correct state of the device to indicate which features
were updated successfully.
To do that we maintain a copy of 'netdev->features' and update it after
successful feature changes, so we can assign it to back to
'netdev->features' if needed.
However, since not all netdev features are handled by the driver (e.g.
GRO/TSO/etc), some features may not be updated correctly in case of an
error updating another feature.
For example, while requesting to disable TSO (feature which is not
handled by the driver) and enable HW-GRO, if an error occurs during
HW-GRO enable, 'oper_features' will be assigned with 'netdev->features'
and HW-GRO turned off. TSO will remain enabled in such case, which is a
bug.
To solve that, instead of using 'netdev->features' as the baseline of
'oper_features' and changing it on set feature success, use 'features'
instead and update it in case of errors.
Fixes: 75b81ce719 ("net/mlx5e: Don't override netdev features field unless in error flow")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Fix a memory leak with decap rule with internal port as destination
device. The driver allocates a modify hdr action but doesn't set
the flow attr modify hdr action which results in skipping releasing
the modify hdr action when releasing the flow.
backtrace:
[<000000005f8c651c>] krealloc+0x83/0xd0
[<000000009f59b143>] alloc_mod_hdr_actions+0x156/0x310 [mlx5_core]
[<000000002257f342>] mlx5e_tc_match_to_reg_set_and_get_id+0x12a/0x360 [mlx5_core]
[<00000000b44ea75a>] mlx5e_tc_add_fdb_flow+0x962/0x1470 [mlx5_core]
[<0000000003e384a0>] __mlx5e_add_fdb_flow+0x54c/0xb90 [mlx5_core]
[<00000000ed8b22b6>] mlx5e_configure_flower+0xe45/0x4af0 [mlx5_core]
[<00000000024f4ab5>] mlx5e_rep_indr_offload.isra.0+0xfe/0x1b0 [mlx5_core]
[<000000006c3bb494>] mlx5e_rep_indr_setup_tc_cb+0x90/0x130 [mlx5_core]
[<00000000d3dac2ea>] tc_setup_cb_add+0x1d2/0x420
Fixes: b16eb3c81f ("net/mlx5: Support internal port as decap route device")
Signed-off-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2021-12-28
This series contains updates to igc driver only.
Vinicius disables support for crosstimestamp on i225-V as lockups are being
observed.
James McLaughlin fixes Tx timestamping support on non-MSI-X platforms.
* '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
igc: Fix TX timestamp support for non-MSI-X platforms
igc: Do not enable crosstimestamping for i225-V models
====================
Link: https://lore.kernel.org/r/20211228182421.340354-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tony Nguyen says:
====================
10GbE Intel Wired LAN Driver Updates 2021-12-28
Alexander Lobakin says:
napi_build_skb() I introduced earlier this year ([0]) aims
to decrease MM pressure and the overhead from in-place
kmem_cache_alloc() on each Rx entry processing by decaching
skbuff_heads from NAPI per-cpu cache filled prior to that by
napi_consume_skb() (so it is sort of a direct shortcut for
free -> mm -> alloc cycle).
Currently, no in-tree drivers use it. Switch all Intel Ethernet
drivers to it to get slight-to-medium perf boosts depending on
the frame size.
ice driver, 50 Gbps link, pktgen + XDP_PASS (local in) sample:
frame_size/nthreads 64/42 128/20 256/8 512/4 1024/2 1532/1
net-next (Kpps) 46062 34654 18248 9830 5343 2714
series 47438 34708 18330 9875 5435 2777
increase 2.9% 0.15% 0.45% 0.46% 1.72% 2.32%
Additionally, e1000's been switched to napi_consume_skb() as it's
safe and works fine there, and there's no point in napi_build_skb()
without paired NAPI cache feeding point.
[0] https://lore.kernel.org/all/20210213141021.87840-1-alobakin@pm.me
* '10GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
ixgbevf: switch to napi_build_skb()
ixgbe: switch to napi_build_skb()
igc: switch to napi_build_skb()
igb: switch to napi_build_skb()
ice: switch to napi_build_skb()
iavf: switch to napi_build_skb()
i40e: switch to napi_build_skb()
e1000: switch to napi_build_skb()
e1000: switch to napi_consume_skb()
====================
Link: https://lore.kernel.org/r/20211228175815.281449-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This patch adds a missing line after the declaration and
fixes the checkpatch warning:
WARNING: Missing a blank line after declarations
+ int desc;
+ for (desc = 0; desc < LTQ_DESC_NUM; desc++)
Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Link: https://lore.kernel.org/r/20211228220031.71576-1-olek2@wp.pl
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
napi_build_skb() reuses per-cpu NAPI skbuff_head cache in order
to save some cycles on freeing/allocating skbuff_heads on every
new Rx or completed Tx.
ixgbevf driver runs Tx completion polling cycle right before the Rx
one and uses napi_consume_skb() to feed the cache with skbuff_heads
of completed entries, so it's never empty and always warm at that
moment. Switch to the napi_build_skb() to relax mm pressure on
heavy Rx.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
napi_build_skb() reuses per-cpu NAPI skbuff_head cache in order
to save some cycles on freeing/allocating skbuff_heads on every
new Rx or completed Tx.
ixgbe driver runs Tx completion polling cycle right before the Rx
one and uses napi_consume_skb() to feed the cache with skbuff_heads
of completed entries, so it's never empty and always warm at that
moment. Switch to the napi_build_skb() to relax mm pressure on
heavy Rx.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
napi_build_skb() reuses per-cpu NAPI skbuff_head cache in order
to save some cycles on freeing/allocating skbuff_heads on every
new Rx or completed Tx.
igc driver runs Tx completion polling cycle right before the Rx
one and uses napi_consume_skb() to feed the cache with skbuff_heads
of completed entries, so it's never empty and always warm at that
moment. Switch to the napi_build_skb() to relax mm pressure on
heavy Rx.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Nechama Kraus <nechamax.kraus@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
napi_build_skb() reuses per-cpu NAPI skbuff_head cache in order
to save some cycles on freeing/allocating skbuff_heads on every
new Rx or completed Tx.
igb driver runs Tx completion polling cycle right before the Rx
one and uses napi_consume_skb() to feed the cache with skbuff_heads
of completed entries, so it's never empty and always warm at that
moment. Switch to the napi_build_skb() to relax mm pressure on
heavy Rx.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
