platform/kernel/linux-rpi.git
3 years agoMerge branch 'net-smc-stats'
David S. Miller [Wed, 16 Jun 2021 19:54:02 +0000 (12:54 -0700)]
Merge branch 'net-smc-stats'

Karsten Graul says:

====================
net/smc: Add SMC statistic support

Please apply the following patch series for smc to netdev's net-next tree.

This v2 is a resend of the code contained in v1 but with an updated
cover letter to describe why we have chosen to use the generic netlink
mechanism to access the smc protocol's statistic data.

The patchset adds statistic support to the SMC protocol. Per-cpu
variables are used to collect the statistic information for better
performance and for reducing concurrency pitfalls. The code that is
collecting statistic data is implemented in macros to increase code
reuse and readability.
The generic netlink mechanism in SMC is extended to provide the
collected statistics to userspace.
Network namespace awareness is also part of the statistics
implementation.

SMC is a protocol interacting with PCI devices (like RoCE Cards) and
runs on top of the TCP protocol. As SMC is a network protocol and not
an ethernet device driver, we decided to use the generic netlink
interface. This should be comparable to what other protocols in the
net subsystem like tipc, ncsi, ieee802154 or tcp, et al, do.
There is already an established internal generic netlink interface
mechanism in SMC which is used to collect SMC Protocol internal
information. This patchset extends that existing mechanism.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet/smc: Make SMC statistics network namespace aware
Guvenc Gulce [Wed, 16 Jun 2021 14:52:58 +0000 (16:52 +0200)]
net/smc: Make SMC statistics network namespace aware

Make the gathered SMC statistics network namespace aware, for each
namespace collect an own set of statistic information.

Signed-off-by: Guvenc Gulce <guvenc@linux.ibm.com>
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet/smc: Add netlink support for SMC fallback statistics
Guvenc Gulce [Wed, 16 Jun 2021 14:52:57 +0000 (16:52 +0200)]
net/smc: Add netlink support for SMC fallback statistics

Add support to collect more detailed SMC fallback reason statistics and
provide these statistics to user space on the netlink interface.

Signed-off-by: Guvenc Gulce <guvenc@linux.ibm.com>
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet/smc: Add netlink support for SMC statistics
Guvenc Gulce [Wed, 16 Jun 2021 14:52:56 +0000 (16:52 +0200)]
net/smc: Add netlink support for SMC statistics

Add the netlink function which collects the statistics information and
delivers it to the userspace.

Signed-off-by: Guvenc Gulce <guvenc@linux.ibm.com>
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet/smc: Add SMC statistics support
Guvenc Gulce [Wed, 16 Jun 2021 14:52:55 +0000 (16:52 +0200)]
net/smc: Add SMC statistics support

Add the ability to collect SMC statistics information. Per-cpu
variables are used to collect the statistic information for better
performance and for reducing concurrency pitfalls. The code that is
collecting statistic data is implemented in macros to increase code
reuse and readability.

Signed-off-by: Guvenc Gulce <guvenc@linux.ibm.com>
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agomlxsw: spectrum_router: remove redundant continue statement
Colin Ian King [Wed, 16 Jun 2021 13:02:58 +0000 (14:02 +0100)]
mlxsw: spectrum_router: remove redundant continue statement

The continue statement at the end of a for-loop has no effect,
remove it.

Addresses-Coverity: ("Continue has no effect")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'nfp-ct-part-two'
David S. Miller [Wed, 16 Jun 2021 19:42:53 +0000 (12:42 -0700)]
Merge branch 'nfp-ct-part-two'

Simon Horman says:

====================
Next set of conntrack patches for the nfp driver

Louis Peens says:

This follows on from the previous series of a similar nature.
Looking at the diagram as explained in the previous series
this implements changes up to the point where the merged
nft entries are saved. There are still bits of stubbed
out code where offloading of the flows will be implemented.

+-------------+                      +----------+
| pre_ct flow +--------+             | nft flow |
+-------------+        v             +------+---+
                  +----------+              |
                  | tc_merge +--------+     |
                  +----------+        v     v
+--------------+       ^           +-------------+
| post_ct flow +-------+       +---+nft_tc merge |
+--------------+               |   +-------------+
                               |
                               |
                               |
                               v
                        Offload to nfp
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: implement action_merge check
Louis Peens [Wed, 16 Jun 2021 10:02:07 +0000 (12:02 +0200)]
nfp: flower-ct: implement action_merge check

Fill in code stub to check that the flow actions are valid for
merge. The actions of the flow X should not conflict with the
matches of flow X+1. For now this check is quite strict and
set_actions are very limited, will need to update this when
NAT support is added.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: fill ct metadata check function
Louis Peens [Wed, 16 Jun 2021 10:02:06 +0000 (12:02 +0200)]
nfp: flower-ct: fill ct metadata check function

Fill in check_meta stub to check that ct_metadata action fields in
the nft flow matches the ct_match data of the post_ct flow.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: fill in ct merge check function
Louis Peens [Wed, 16 Jun 2021 10:02:05 +0000 (12:02 +0200)]
nfp: flower-ct: fill in ct merge check function

Replace merge check stub code with the actual implementation. This
checks that the match parts of two tc flows does not conflict.
Only overlapping keys needs to be checked, and only the narrowest
masked parts needs to be checked, so each key is masked with the
AND'd result of both masks before comparing.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: implement code to save merge of tc and nft flows
Louis Peens [Wed, 16 Jun 2021 10:02:04 +0000 (12:02 +0200)]
nfp: flower-ct: implement code to save merge of tc and nft flows

Add in the code to merge the tc_merge objects with the flows
received from nft. At the moment flows are just merged blindly
as the validity check functions are stubbed out, this will
be populated in follow-up patches.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: add nft_merge table
Louis Peens [Wed, 16 Jun 2021 10:02:03 +0000 (12:02 +0200)]
nfp: flower-ct: add nft_merge table

Add table and struct to save the result of the three-way merge
between pre_ct,post_ct, and nft flows. Merging code is to be
added in follow-up patches.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: make a full copy of the rule when it is a NFT flow
Yinjun Zhang [Wed, 16 Jun 2021 10:02:02 +0000 (12:02 +0200)]
nfp: flower-ct: make a full copy of the rule when it is a NFT flow

The nft flow will be destroyed after offload cb returns. This means
we need save a full copy of it since it can be referenced through
other paths other than just the offload cb, for example when a new
pre_ct or post_ct entry is added, and it needs to be merged with
an existing nft entry.

Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: add nft flows to nft list
Louis Peens [Wed, 16 Jun 2021 10:02:01 +0000 (12:02 +0200)]
nfp: flower-ct: add nft flows to nft list

Implement code to add and remove nft flows to the relevant list.
Registering and deregistering the callback function for the nft
table is quite complicated. The safest is to delete the callback
on the removal of the last pre_ct flow. This is because if this
is also the latest pre_ct flow in software it means that this
specific nft table will be freed, so there will not be a later
opportunity to do this. Another place where it looks possible
to delete the callback is when the last nft_flow is deleted,
but this happens under the flow_table lock, which is also taken
when deregistering the callback, leading to a deadlock situation.

This means the final solution here is to delete the callback
when removing the last pre_ct flow, and then clean up any
remaining nft_flow entries which may still be present, since
there will never be a callback now to do this, leaving them
orphaned if not cleaned up here as well.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: add nft callback stubs
Louis Peens [Wed, 16 Jun 2021 10:02:00 +0000 (12:02 +0200)]
nfp: flower-ct: add nft callback stubs

Add register/unregister of the nft callback. For now just add
stub code to accept the flows, but don't do anything with it.
Decided to accept the flows since netfilter will keep on trying
to offload a flow if it was rejected, which is quite noisy.
Follow-up patches will start implementing the functions to add
nft flows to the relevant tables.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonfp: flower-ct: add delete flow handling for ct
Louis Peens [Wed, 16 Jun 2021 10:01:59 +0000 (12:01 +0200)]
nfp: flower-ct: add delete flow handling for ct

Add functions to handle delete flow callbacks for ct flows. Also
accept the flows for offloading by returning 0 instead of -EOPNOTSUPP.
Flows will still not actually be offloaded to hw, but at this point
it's difficult to not accept the flows and also exercise the cleanup
paths properly. Traffic will still be handled safely through the
fallback path.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'net-phy-cleanups'
David S. Miller [Wed, 16 Jun 2021 19:34:08 +0000 (12:34 -0700)]
Merge branch 'net-phy-cleanups'

Weihang Li says:

====================
net: phy: fix some coding-style issues

Make some cleanups according to the coding style of kernel.

Changes since v1:
- Update commit description of #1 and #3.
- Avoid changing the indentation in #2.
- Change a group of if-else statement into switch from #4 and put it into
  a single patch.
- Put '|' at the end of line in #5 and #7.
- Avoid deleting spaces in definition of 'settings' in #5.
- Drop #8 from the series which needs more discussion with David.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: phy: replace if-else statements with switch
Weihang Li [Wed, 16 Jun 2021 10:01:26 +0000 (18:01 +0800)]
net: phy: replace if-else statements with switch

Switch statement is clearer than a group of 'if-else'.

Signed-off-by: Weihang Li <liweihang@huawei.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: phy: remove unnecessary line continuation
Wenpeng Liang [Wed, 16 Jun 2021 10:01:25 +0000 (18:01 +0800)]
net: phy: remove unnecessary line continuation

Avoid unnecessary line continuations, and put '|' at the end of line.

Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: phy: print the function name by __func__ instead of an fixed string
Wenpeng Liang [Wed, 16 Jun 2021 10:01:24 +0000 (18:01 +0800)]
net: phy: print the function name by __func__ instead of an fixed string

It's better to use __func__ than a fixed string to print a function's
name.

Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: phy: fix formatting issues with braces
Wenpeng Liang [Wed, 16 Jun 2021 10:01:23 +0000 (18:01 +0800)]
net: phy: fix formatting issues with braces

Fix following format issues:
1. open brace '{' following function definitions should go to the next
   line.
2. braces {} are not necessary for single line statements.
3. else should follow close brace '}'.

Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: phy: fix space alignment issues
Wenpeng Liang [Wed, 16 Jun 2021 10:01:22 +0000 (18:01 +0800)]
net: phy: fix space alignment issues

There are some space related issues, including spaces at the start of the
line, before tabs, after open parenthesis and before close parenthesis.

Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: phy: delete repeated words of comments
Wenpeng Liang [Wed, 16 Jun 2021 10:01:21 +0000 (18:01 +0800)]
net: phy: delete repeated words of comments

There are some repeated words in some comments, they should be deleted.

Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: phy: correct format of block comments
Wenpeng Liang [Wed, 16 Jun 2021 10:01:20 +0000 (18:01 +0800)]
net: phy: correct format of block comments

Block comments should not use a trailing */ on a separate line and every
line of a block comment should start with an '*'.

Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: phy: change format of some declarations
Wenpeng Liang [Wed, 16 Jun 2021 10:01:19 +0000 (18:01 +0800)]
net: phy: change format of some declarations

Add a blank line after declarations, change the order of them and put the
assignments and declarations together.

Cc: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: qualcomm: rmnet: Remove some unneeded casts
Subash Abhinov Kasiviswanathan [Wed, 16 Jun 2021 07:59:13 +0000 (01:59 -0600)]
net: qualcomm: rmnet: Remove some unneeded casts

Remove the explicit casts in the checksum complement functions
and pass the actual protocol specific headers instead.

Signed-off-by: Subash Abhinov Kasiviswanathan <subashab@codeaurora.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: iosm: remove the repeated declaration and comment
Shaokun Zhang [Wed, 16 Jun 2021 07:25:40 +0000 (15:25 +0800)]
net: iosm: remove the repeated declaration and comment

Function 'ipc_mmio_get_cp_version' is declared twice, so remove the
repeated declaration and wrong comments.

Cc: M Chetan Kumar <m.chetan.kumar@intel.com>
Cc: Intel Corporation <linuxwwan@intel.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
Reviewed-by: M Chetan Kumar <m.chetan.kumar@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: xrs700x: forward HSR supervision frames
George McCollister [Wed, 16 Jun 2021 01:39:03 +0000 (20:39 -0500)]
net: dsa: xrs700x: forward HSR supervision frames

Forward supervision frames between redunant HSR ports. This was broken
in the last commit.

Fixes: 1a42624aecba ("net: dsa: xrs700x: allow HSR/PRP supervision dupes for node_table")
Signed-off-by: George McCollister <george.mccollister@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: qualcomm: rmnet: Allow partial updates of IFLA_FLAGS
Bjorn Andersson [Tue, 15 Jun 2021 23:27:07 +0000 (18:27 -0500)]
net: qualcomm: rmnet: Allow partial updates of IFLA_FLAGS

The idiomatic way to handle the changelink flags/mask pair seems to be
allow partial updates of the driver's link flags. In contrast the rmnet
driver masks the incoming flags and then use that as the new flags.

Change the rmnet driver to follow the common scheme, before the
introduction of IFLA_RMNET_FLAGS handling in iproute2 et al.

Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Reviewed-by: Alex Elder <elder@linaro.org>
Reviewed-by: Subash Abhinov Kasiviswanathan <subashab@codeaurora.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hsr: don't check sequence number if tag removal is offloaded
George McCollister [Tue, 15 Jun 2021 17:50:37 +0000 (12:50 -0500)]
net: hsr: don't check sequence number if tag removal is offloaded

Don't check the sequence number when deciding when to update time_in in
the node table if tag removal is offloaded since the sequence number is
part of the tag. This fixes a problem where the times in the node table
wouldn't update when 0 appeared to be before or equal to seq_out when
tag removal was offloaded.

Signed-off-by: George McCollister <george.mccollister@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: stmmac: Fix error return code in ingenic_mac_probe()
Wei Yongjun [Tue, 15 Jun 2021 17:21:55 +0000 (17:21 +0000)]
net: stmmac: Fix error return code in ingenic_mac_probe()

Fix to return a negative error code from the error handling
case instead of 0, as done elsewhere in this function.

Fixes: 2bb4b98b60d7 ("net: stmmac: Add Ingenic SoCs MAC support.")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: chelsio: cxgb4: use eth_zero_addr() to assign zero address
Yang Yingliang [Wed, 16 Jun 2021 06:43:18 +0000 (14:43 +0800)]
net: chelsio: cxgb4: use eth_zero_addr() to assign zero address

Using eth_zero_addr() to assign zero address insetad of
inefficient copy from an array.

Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'cosa-cleanups'
David S. Miller [Wed, 16 Jun 2021 07:52:34 +0000 (00:52 -0700)]
Merge branch 'cosa-cleanups'

Peng Li says:

====================
net: cosa: clean up some code style issues

This patchset clean up some code style issues.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: remove redundant spaces
Peng Li [Wed, 16 Jun 2021 07:23:41 +0000 (15:23 +0800)]
net: cosa: remove redundant spaces

According to the chackpatch.pl,
no spaces is necessary at the start of a line,
no space is necessary after a cast.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: remove trailing whitespaces
Peng Li [Wed, 16 Jun 2021 07:23:40 +0000 (15:23 +0800)]
net: cosa: remove trailing whitespaces

This patch removes trailing whitespaces.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: add some required spaces
Peng Li [Wed, 16 Jun 2021 07:23:39 +0000 (15:23 +0800)]
net: cosa: add some required spaces

Add space required before the open parenthesis '(' and '{'.
Add space required after that close brace '}' and ','
Add spaces required around that '=' , '&', '*', '|', '+', '/' and '-'.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: fix the code style issue about trailing statements
Peng Li [Wed, 16 Jun 2021 07:23:38 +0000 (15:23 +0800)]
net: cosa: fix the code style issue about trailing statements

Trailing statements should be on next line.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: fix the alignment issue
Peng Li [Wed, 16 Jun 2021 07:23:37 +0000 (15:23 +0800)]
net: cosa: fix the alignment issue

Alignment should match open parenthesis.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: use BIT macro
Peng Li [Wed, 16 Jun 2021 07:23:36 +0000 (15:23 +0800)]
net: cosa: use BIT macro

This patch uses the BIT macro for setting individual bits,
to fix the following checkpatch.pl issue:
CHECK: Prefer using the BIT macro.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: add necessary () to macro argument
Peng Li [Wed, 16 Jun 2021 07:23:35 +0000 (15:23 +0800)]
net: cosa: add necessary () to macro argument

Macro argument 'cosa' may be better as '(cosa)' to avoid
precedence issues.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: remove redundant braces {}
Peng Li [Wed, 16 Jun 2021 07:23:34 +0000 (15:23 +0800)]
net: cosa: remove redundant braces {}

This patch removes redundant braces {}, to fix the
checkpatch.pl warning:
"braces {} are not necessary for single statement blocks".

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: add braces {} to all arms of the statement
Peng Li [Wed, 16 Jun 2021 07:23:33 +0000 (15:23 +0800)]
net: cosa: add braces {} to all arms of the statement

Braces {} should be used on all arms of this statement.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: fix the comments style issue
Peng Li [Wed, 16 Jun 2021 07:23:32 +0000 (15:23 +0800)]
net: cosa: fix the comments style issue

Networking block comments don't use an empty /* line,
use /* Comment...

Block comments use * on subsequent lines.
Block comments use a trailing */ on a separate line.

This patch fixes the comments style issues.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: move out assignment in if condition
Peng Li [Wed, 16 Jun 2021 07:23:31 +0000 (15:23 +0800)]
net: cosa: move out assignment in if condition

Should not use assignment in if condition.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: replace comparison to NULL with "!chan->rx_skb"
Peng Li [Wed, 16 Jun 2021 07:23:30 +0000 (15:23 +0800)]
net: cosa: replace comparison to NULL with "!chan->rx_skb"

According to the chackpatch.pl, comparison to NULL could
be written "!chan->rx_skb".

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: fix the code style issue about "foo* bar"
Peng Li [Wed, 16 Jun 2021 07:23:29 +0000 (15:23 +0800)]
net: cosa: fix the code style issue about "foo* bar"

Fix the checkpatch error as "foo* bar" should be "foo *bar".

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: add blank line after declarations
Peng Li [Wed, 16 Jun 2021 07:23:28 +0000 (15:23 +0800)]
net: cosa: add blank line after declarations

This patch fixes the checkpatch error about missing a blank line
after declarations.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: cosa: remove redundant blank lines
Peng Li [Wed, 16 Jun 2021 07:23:27 +0000 (15:23 +0800)]
net: cosa: remove redundant blank lines

This patch removes some redundant blank lines.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: iosm: add missing MODULE_DEVICE_TABLE
Zou Wei [Wed, 16 Jun 2021 04:07:27 +0000 (12:07 +0800)]
net: iosm: add missing MODULE_DEVICE_TABLE

This patch adds missing MODULE_DEVICE_TABLE definition which generates
correct modalias for automatic loading of this driver when it is built
as an external module.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zou Wei <zou_wei@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoqlcnic: Use list_for_each_entry() to simplify code in qlcnic_main.c
Wang Hai [Wed, 16 Jun 2021 04:21:06 +0000 (12:21 +0800)]
qlcnic: Use list_for_each_entry() to simplify code in qlcnic_main.c

Convert list_for_each() to list_for_each_entry() where
applicable. This simplifies the code.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wang Hai <wanghai38@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoethtool: add a stricter length check
Jakub Kicinski [Wed, 16 Jun 2021 03:33:38 +0000 (20:33 -0700)]
ethtool: add a stricter length check

There has been a few errors in the ethtool reply size calculations,
most of those are hard to trigger during basic testing because of
skb size rounding up and netdev names being shorter than max.
Add a more precise check.

This change will affect the value of payload length displayed in
case of -EMSGSIZE but that should be okay, "payload length" isn't
a well defined term here.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoinet_diag: add support for tw_mark
Maciej Żenczykowski [Wed, 16 Jun 2021 06:06:04 +0000 (23:06 -0700)]
inet_diag: add support for tw_mark

Timewait sockets have included mark since approx 4.18.

Cc: Eric Dumazet <edumazet@google.com>
Cc: Jon Maxwell <jmaxwell37@gmail.com>
Fixes: 00483690552c ("tcp: Add mark for TIMEWAIT sockets")
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jon Maxwell <jmaxwell37@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: mhi_net: make mhi_wwan_ops static
Jiapeng Chong [Wed, 16 Jun 2021 05:53:09 +0000 (13:53 +0800)]
net: mhi_net: make mhi_wwan_ops static

This symbol is not used outside of net.c, so marks it static.

Fix the following sparse warning:

drivers/net/mhi/net.c:385:23: warning: symbol 'mhi_wwan_ops' was not
declared. Should it be static?

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'hns3-next'
David S. Miller [Wed, 16 Jun 2021 07:36:07 +0000 (00:36 -0700)]
Merge branch 'hns3-next'

Guangbin Huang says:

====================
net: hns3: updates for -next

This series includes some optimization in IO path for the HNS3 ethernet
driver.
====================

Cc: Loic Poulain <loic.poulain@linaro.org>
Cc: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: use bounce buffer when rx page can not be reused
Yunsheng Lin [Wed, 16 Jun 2021 06:36:17 +0000 (14:36 +0800)]
net: hns3: use bounce buffer when rx page can not be reused

Currently rx page will be reused to receive future packet when
the stack releases the previous skb quickly. If the old page
can not be reused, a new page will be allocated and mapped,
which comsumes a lot of cpu when IOMMU is in the strict mode,
especially when the application and irq/NAPI happens to run on
the same cpu.

So allocate a new frag to memcpy the data to avoid the costly
IOMMU unmapping/mapping operation, and add "frag_alloc_err"
and "frag_alloc" stats in "ethtool -S ethX" cmd.

The throughput improves above 50% when running single thread of
iperf using TCP when IOMMU is in strict mode and iperf shares the
same cpu with irq/NAPI(rx_copybreak = 2048 and mtu = 1500).

Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: optimize the rx page reuse handling process
Yunsheng Lin [Wed, 16 Jun 2021 06:36:16 +0000 (14:36 +0800)]
net: hns3: optimize the rx page reuse handling process

Current rx page offset only reset to zero when all the below
conditions are satisfied:
1. rx page is only owned by driver.
2. rx page is reusable.
3. the page offset that is above to be given to the stack has
reached the end of the page.

If the page offset is over the hns3_buf_size(), it means the
buffer below the offset of the page is usable when the above
condition 1 & 2 are satisfied, so page offset can be reset to
zero instead of increasing the offset. We may be able to always
reuse the first 4K buffer of a 64K page, which means we can
limit the hot buffer size as much as possible.

The above optimization is a side effect when refacting the
rx page reuse handling in order to support the rx copybreak.

Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: support dma_map_sg() for multi frags skb
Yunsheng Lin [Wed, 16 Jun 2021 06:36:15 +0000 (14:36 +0800)]
net: hns3: support dma_map_sg() for multi frags skb

Using the queue based tx buffer, it is also possible to allocate a
sgl buffer, and use skb_to_sgvec() to convert the skb to the sgvec
in order to support the dma_map_sg() to decreases the overhead of
IOMMU mapping and unmapping.

Firstly, it reduces the number of buffers. For example, a tcp skb
may have a 66-byte header and 3 fragments of 4328, 32768, and 28064
bytes. With this patch, dma_map_sg() will combine them into two
buffers, 66-bytes header and one 65160-bytes fragment by using IOMMU.

Secondly, it reduces the number of dma mapping and unmapping. All the
original 4 buffers are mapped only once rather than 4 times.

The throughput improves above 10% when running single thread of iperf
using TCP when IOMMU is in strict mode.

Suggested-by: Barry Song <song.bao.hua@hisilicon.com>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: add support to query tx spare buffer size for pf
Huazhong Tan [Wed, 16 Jun 2021 06:36:14 +0000 (14:36 +0800)]
net: hns3: add support to query tx spare buffer size for pf

Add support to query tx spare buffer size from configuration
file, and use this info to do spare buffer initialization when
the module parameter 'tx_spare_buf_size' is not specified.

Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: use tx bounce buffer for small packets
Yunsheng Lin [Wed, 16 Jun 2021 06:36:13 +0000 (14:36 +0800)]
net: hns3: use tx bounce buffer for small packets

when the packet or frag size is small, it causes both security and
performance issue. As dma can't map sub-page, this means some extra
kernel data is visible to devices. On the other hand, the overhead
of dma map and unmap is huge when IOMMU is on.

So add a queue based tx shared bounce buffer to memcpy the small
packet when the len of the xmitted skb is below tx_copybreak.
Add tx_spare_buf_size module param to set the size of tx spare
buffer, and add set/get_tunable to set or query the tx_copybreak.

The throughtput improves from 30 Gbps to 90+ Gbps when running 16
netperf threads with 32KB UDP message size when IOMMU is in the
strict mode(tx_copybreak = 2000 and mtu = 1500).

Suggested-by: Barry Song <song.bao.hua@hisilicon.com>
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: refactor for hns3_fill_desc() function
Yunsheng Lin [Wed, 16 Jun 2021 06:36:12 +0000 (14:36 +0800)]
net: hns3: refactor for hns3_fill_desc() function

Factor out hns3_fill_desc() so that it can be reused in the
tx bounce supporting.

Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: minor refactor related to desc_cb handling
Yunsheng Lin [Wed, 16 Jun 2021 06:36:11 +0000 (14:36 +0800)]
net: hns3: minor refactor related to desc_cb handling

desc_cb is used to store mapping and freeing info for the
corresponding desc, which is used in the cleaning process.
There will be more desc_cb type coming up when supporting the
tx bounce buffer, change desc_cb type to bit-wise value in order
to reduce the desc_cb type checking operation in the data path.

Also move the desc_cb type definition to hns3_enet.h because it
is only used in hns3_enet.c, and declare a local variable desc_cb
in hns3_clear_desc() to reduce lines of code.

Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ti: add pp skb recycling support
Lorenzo Bianconi [Tue, 15 Jun 2021 13:27:41 +0000 (15:27 +0200)]
net: ti: add pp skb recycling support

As already done for mvneta and mvpp2, enable skb recycling for ti
ethernet drivers

ti driver on net-next:
----------------------
[perf top]
 47.15%  [kernel]     [k] _raw_spin_unlock_irqrestore
 11.77%  [kernel]     [k] __cpdma_chan_free
  3.16%  [kernel]     [k] ___bpf_prog_run
  2.52%  [kernel]     [k] cpsw_rx_vlan_encap
  2.34%  [kernel]     [k] __netif_receive_skb_core
  2.27%  [kernel]     [k] free_unref_page
  2.26%  [kernel]     [k] kmem_cache_free
  2.24%  [kernel]     [k] kmem_cache_alloc
  1.69%  [kernel]     [k] __softirqentry_text_start
  1.61%  [kernel]     [k] cpsw_rx_handler
  1.19%  [kernel]     [k] page_pool_release_page
  1.19%  [kernel]     [k] clear_bits_ll
  1.15%  [kernel]     [k] page_frag_free
  1.06%  [kernel]     [k] __dma_page_dev_to_cpu
  0.99%  [kernel]     [k] memset
  0.94%  [kernel]     [k] __alloc_pages_bulk
  0.92%  [kernel]     [k] kfree_skb
  0.85%  [kernel]     [k] packet_rcv
  0.78%  [kernel]     [k] page_address
  0.75%  [kernel]     [k] v7_dma_inv_range
  0.71%  [kernel]     [k] __lock_text_start

[iperf3 tcp]
[  5]   0.00-10.00  sec   873 MBytes   732 Mbits/sec    0   sender
[  5]   0.00-10.01  sec   866 MBytes   726 Mbits/sec        receiver

ti + skb recycling:
-------------------
[perf top]
 40.58%  [kernel]    [k] _raw_spin_unlock_irqrestore
 16.18%  [kernel]    [k] __softirqentry_text_start
 10.33%  [kernel]    [k] __cpdma_chan_free
  2.62%  [kernel]    [k] ___bpf_prog_run
  2.05%  [kernel]    [k] cpsw_rx_vlan_encap
  2.00%  [kernel]    [k] kmem_cache_alloc
  1.86%  [kernel]    [k] __netif_receive_skb_core
  1.80%  [kernel]    [k] kmem_cache_free
  1.63%  [kernel]    [k] cpsw_rx_handler
  1.12%  [kernel]    [k] cpsw_rx_mq_poll
  1.11%  [kernel]    [k] page_pool_put_page
  1.04%  [kernel]    [k] _raw_spin_unlock
  0.97%  [kernel]    [k] clear_bits_ll
  0.90%  [kernel]    [k] packet_rcv
  0.88%  [kernel]    [k] __dma_page_dev_to_cpu
  0.85%  [kernel]    [k] kfree_skb
  0.80%  [kernel]    [k] memset
  0.71%  [kernel]    [k] __lock_text_start
  0.66%  [kernel]    [k] v7_dma_inv_range
  0.64%  [kernel]    [k] gen_pool_free_owner

[iperf3 tcp]
[  5]   0.00-10.00  sec   884 MBytes   742 Mbits/sec    0   sender
[  5]   0.00-10.01  sec   878 MBytes   735 Mbits/sec        receiver

Tested-by: Grygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: wwan: iosm: Fix htmldocs warnings
M Chetan Kumar [Tue, 15 Jun 2021 13:08:22 +0000 (18:38 +0530)]
net: wwan: iosm: Fix htmldocs warnings

Fixes .rst file warnings seen on linux-next build.

Fixes: f7af616c632e ("net: iosm: infrastructure")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: M Chetan Kumar <m.chetan.kumar@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoocteontx2-pf: Fix spelling mistake "morethan" -> "more than"
Colin Ian King [Tue, 15 Jun 2021 10:14:57 +0000 (11:14 +0100)]
octeontx2-pf: Fix spelling mistake "morethan" -> "more than"

There is a spelling mistake in a dev_err message. Fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: b53: remove redundant null check on dev
Colin Ian King [Tue, 15 Jun 2021 09:05:16 +0000 (10:05 +0100)]
net: dsa: b53: remove redundant null check on dev

The pointer dev can never be null, the null check is redundant
and can be removed. Cleans up a static analysis warning that
pointer priv is dereferencing dev before dev is being null
checked.

Addresses-Coverity: ("Dereference before null check")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bonding: Use per-cpu rr_tx_counter
Jussi Maki [Tue, 15 Jun 2021 08:54:15 +0000 (08:54 +0000)]
net: bonding: Use per-cpu rr_tx_counter

The round-robin rr_tx_counter was shared across CPUs leading to
significant cache thrashing at high packet rates. This patch switches
the round-robin packet counter to use a per-cpu variable to decide
the destination slave.

On a test with 2x100Gbit ICE nic with pktgen_sample_04_many_flows.sh
(-s 64 -t 32) the tx rate was 19.6Mpps before and 22.3Mpps after
this patch.

"perf top -e cache_misses" before:
    12.31%  [bonding]       [k] bond_xmit_roundrobin_slave_get
    10.59%  [sch_fq_codel]  [k] fq_codel_dequeue
     9.34%  [kernel]        [k] skb_release_data
after:
    15.42%  [sch_fq_codel]  [k] fq_codel_dequeue
    10.06%  [kernel]        [k] __memset
     9.12%  [kernel]        [k] skb_release_data

Signed-off-by: Jussi Maki <joamaki@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonetlabel: Fix memory leak in netlbl_mgmt_add_common
Liu Shixin [Tue, 15 Jun 2021 02:14:44 +0000 (10:14 +0800)]
netlabel: Fix memory leak in netlbl_mgmt_add_common

Hulk Robot reported memory leak in netlbl_mgmt_add_common.
The problem is non-freed map in case of netlbl_domhsh_add() failed.

BUG: memory leak
unreferenced object 0xffff888100ab7080 (size 96):
  comm "syz-executor537", pid 360, jiffies 4294862456 (age 22.678s)
  hex dump (first 32 bytes):
    05 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    fe 00 00 00 00 00 00 00 00 00 00 00 00 00 00 01  ................
  backtrace:
    [<0000000008b40026>] netlbl_mgmt_add_common.isra.0+0xb2a/0x1b40
    [<000000003be10950>] netlbl_mgmt_add+0x271/0x3c0
    [<00000000c70487ed>] genl_family_rcv_msg_doit.isra.0+0x20e/0x320
    [<000000001f2ff614>] genl_rcv_msg+0x2bf/0x4f0
    [<0000000089045792>] netlink_rcv_skb+0x134/0x3d0
    [<0000000020e96fdd>] genl_rcv+0x24/0x40
    [<0000000042810c66>] netlink_unicast+0x4a0/0x6a0
    [<000000002e1659f0>] netlink_sendmsg+0x789/0xc70
    [<000000006e43415f>] sock_sendmsg+0x139/0x170
    [<00000000680a73d7>] ____sys_sendmsg+0x658/0x7d0
    [<0000000065cbb8af>] ___sys_sendmsg+0xf8/0x170
    [<0000000019932b6c>] __sys_sendmsg+0xd3/0x190
    [<00000000643ac172>] do_syscall_64+0x37/0x90
    [<000000009b79d6dc>] entry_SYSCALL_64_after_hwframe+0x44/0xae

Fixes: 63c416887437 ("netlabel: Add network address selectors to the NetLabel/LSM domain mapping")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Liu Shixin <liushixin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge tag 'mlx5-updates-2021-06-14' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Tue, 15 Jun 2021 18:14:21 +0000 (11:14 -0700)]
Merge tag 'mlx5-updates-2021-06-14' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5-updates-2021-06-14

1) Trivial Lag refactroing in preparation for upcomming Single FDB lag feature
 - First 3 patches

2) Scalable IRQ distriburion for Sub-functions

A subfunction (SF) is a lightweight function that has a parent PCI
function (PF) on which it is deployed.

Currently, mlx5 subfunction is sharing the IRQs (MSI-X) with their
parent PCI function.

Before this series the PF allocates enough IRQs to cover
all the cores in a system, Newly created SFs will re-use all the IRQs
that the PF has allocated for itself.
Hence, the more SFs are created, there are more EQs per IRQs. Therefore,
whenever we handle an interrupt, we need to pull all SFs EQs and PF EQs
instead of PF EQs without SFs on the system. This leads to a hard impact
on the performance of SFs and PF.

For example, on machine with:
Intel(R) Xeon(R) CPU E5-2697 v3 @ 2.60GHz with 56 cores.
PCI Express 3 with BW of 126 Gb/s.
ConnectX-5 Ex; EDR IB (100Gb/s) and 100GbE; dual-port QSFP28; PCIe4.0 x16.

test case: iperf TX BW single CPU, affinity of app and IRQ are the same.
PF only: no SFs on the system, 56 IRQs.
SF (before), 250 SFs Sharing the same 56 IRQs .
SF (now),    250 SFs + 255 avaiable IRQs for the NIC. (please see IRQ spread scheme below).

    application SF-IRQ  channel   BW(Gb/sec)         interrupts/sec
            iperf TX            affinity
PF only     cpu={0}     cpu={0} cpu={0}   79                 8200
SF (before) cpu={0}     cpu={0} cpu={0}   51.3 (-35%)        9500
SF (now)    cpu={0}     cpu={0} cpu={0}   78 (-2%)           8200

command:
$ taskset -c 0 iperf -c 11.1.1.1 -P 3 -i 6 -t 30 | grep SUM

The different between the SF examples is that before this series we
allocate num_cpus (56) IRQs, and all of them were shared among the PF
and the SFs. And after this series, we allocate 255 IRQs, and we spread
the SFs among the above IRQs. This have significantly decreased the load
on each IRQ and the number of EQs per IRQ is down by 95% (251->11).

In this patchset the solution proposed is to have a dedicated IRQ pool
for SFs to use. the pool will allocate a large number of IRQs
for SFs to grab from in order to minimize irq sharing between the
different SFs.
IRQs will not be requested from the OS until they are 1st requested by
an SF consumer, and will be eventually released when the last SF consumer
releases them.

For the detailed IRQ spread and allocation scheme  please see last patch:
("net/mlx5: Round-Robin EQs over IRQs")
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'occteontx2-rate-limit-offload'
David S. Miller [Tue, 15 Jun 2021 18:11:05 +0000 (11:11 -0700)]
Merge branch 'occteontx2-rate-limit-offload'

Subbaraya Sundeep says:

====================
octeontx2: Add ingress ratelimit offload

This patchset adds ingress rate limiting hardware
offload support for CN10K silicons. Police actions
are added for TC matchall and flower filters.
CN10K has ingress rate limiting feature where
a receive queue is mapped to bandwidth profile
and the profile is configured with rate and burst
parameters by software. CN10K hardware supports
three levels of ingress policing or ratelimiting.
Multiple leaf profiles can  point to a single mid
level profile and multiple mid level profile can
point to a single top level one. Only leaf level
profiles are used for configuring rate limiting.

Patch 1 adds the new bandwidth profile contexts
in AF driver similar to other hardware contexts
Patch 2 adds the debugfs changes to dump bandwidth
profile contexts
Patch 3 adds support for police action with TC matchall filter
Patch 4 uses NL_SET_ERR_MSG_MOD for tc code
Patch 5 adds support for police action with TC flower filter
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoocteontx2-pf: Add police action for TC flower
Subbaraya Sundeep [Tue, 15 Jun 2021 11:34:31 +0000 (17:04 +0530)]
octeontx2-pf: Add police action for TC flower

Added police action for ingress TC flower
hardware offload. With this rate limiting can be
done per flow. Since rate limiting is tied to
RQs in hardware the number of TC flower filters
with action as police is limited to number
of receive queues of the interface. Both bps
and pps modes are supported.

Examples to rate limit a flow:
$ ethtool -K eth0 hw-tc-offload on
$ tc qdisc add dev eth0 ingress
$ tc filter add dev eth0 parent ffff: protocol ip \
  flower ip_proto udp dst_port 80 action \
  police rate 100Mbit burst 32Kbit

$ tc filter add dev eth0 parent ffff: \
  protocol ip flower dst_mac 5e:b2:34:ee:29:49 \
  action police pkts_rate 5000 pkts_burst 2048

Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoocteontx2-pf: Use NL_SET_ERR_MSG_MOD for TC
Subbaraya Sundeep [Tue, 15 Jun 2021 11:34:30 +0000 (17:04 +0530)]
octeontx2-pf: Use NL_SET_ERR_MSG_MOD for TC

This patch modifies all netdev_err messages in
tc code to NL_SET_ERR_MSG_MOD. NL_SET_ERR_MSG_MOD
does not support format specifiers yet hence
netdev_err messages with only strings are modified.

Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoocteontx2-pf: TC_MATCHALL ingress ratelimiting offload
Sunil Goutham [Tue, 15 Jun 2021 11:34:29 +0000 (17:04 +0530)]
octeontx2-pf: TC_MATCHALL ingress ratelimiting offload

Add TC_MATCHALL ingress ratelimiting offload support with POLICE
action for entire traffic coming into the interface.

Eg: To ratelimit ingress traffic to 100Mbps

$ ethtool -K eth0 hw-tc-offload on
$ tc qdisc add dev eth0 clsact
$ tc filter add dev eth0 ingress matchall skip_sw \
                action police rate 100Mbit burst 32Kbit

To support this, a leaf level bandwidth profile is allocated and all
RQs' contexts used by this interface are updated to point to it.
And the leaf level bandwidth profile is configured with user specified
rate and burst sizes.

Co-developed-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoocteontx2-af: cn10k: Debugfs support for bandwidth profiles
Sunil Goutham [Tue, 15 Jun 2021 11:34:28 +0000 (17:04 +0530)]
octeontx2-af: cn10k: Debugfs support for bandwidth profiles

Added support for dumping current resource status of bandwidth
profiles and contexts of allocated profiles via debugfs.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoocteontx2-af: cn10k: Bandwidth profiles config support
Sunil Goutham [Tue, 15 Jun 2021 11:34:27 +0000 (17:04 +0530)]
octeontx2-af: cn10k: Bandwidth profiles config support

CN10K silicons supports hierarchial ingress packet ratelimiting.
There are 3 levels of profilers supported leaf, mid and top.
Ratelimiting is done after packet forwarding decision is taken
and a NIXLF's RQ is identified to DMA the packet. RQ's context
points to a leaf bandwidth profile which can be configured
to achieve desired ratelimit.

This patch adds logic for management of these bandwidth profiles
ie profile alloc, free, context update etc.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'pci200syn-cleanups'
David S. Miller [Tue, 15 Jun 2021 18:03:17 +0000 (11:03 -0700)]
Merge branch 'pci200syn-cleanups'

Peng Li says:

====================
net: pci200syn: clean up some code style issues

This patchset clean up some code style issues.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: pci200syn: fix the comments style issue
Peng Li [Tue, 15 Jun 2021 13:54:23 +0000 (21:54 +0800)]
net: pci200syn: fix the comments style issue

Networking block comments don't use an empty /* line,
use /* Comment...

This patch fixes the comments style issues.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: pci200syn: add necessary () to macro argument
Peng Li [Tue, 15 Jun 2021 13:54:22 +0000 (21:54 +0800)]
net: pci200syn: add necessary () to macro argument

Macro argument 'card' may be better as '(card)' to
avoid precedence issues.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: pci200syn: add some required spaces
Peng Li [Tue, 15 Jun 2021 13:54:21 +0000 (21:54 +0800)]
net: pci200syn: add some required spaces

Add spaces required after that close brace '}'.
Add spaces required before the open parenthesis '('.
Add spaces required after that ','.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: pci200syn: replace comparison to NULL with "!card"
Peng Li [Tue, 15 Jun 2021 13:54:20 +0000 (21:54 +0800)]
net: pci200syn: replace comparison to NULL with "!card"

According to the chackpatch.pl, comparison to NULL could
be written "!card".

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: pci200syn: add blank line after declarations
Peng Li [Tue, 15 Jun 2021 13:54:19 +0000 (21:54 +0800)]
net: pci200syn: add blank line after declarations

This patch fixes the checkpatch error about missing a blank line
after declarations.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: pci200syn: remove redundant blank lines
Peng Li [Tue, 15 Jun 2021 13:54:18 +0000 (21:54 +0800)]
net: pci200syn: remove redundant blank lines

This patch removes some redundant blank lines.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'z85230-cleanups'
David S. Miller [Tue, 15 Jun 2021 17:55:18 +0000 (10:55 -0700)]
Merge branch 'z85230-cleanups'

Peng Li says:

====================
net: z85230: clean up some code style issues

This patchset clean up some code style issues.

---
Change Log:
V1 -> V2:
1, fix the comments from Andrew, add commit message to [patch 04/11]
   about remove volatile.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: remove unnecessary out of memory message
Peng Li [Tue, 15 Jun 2021 02:43:45 +0000 (10:43 +0800)]
net: z85230: remove unnecessary out of memory message

This patch removes unnecessary out of memory message,
to fix the following checkpatch.pl warning:
"WARNING: Possible unnecessary 'out of memory' message"

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: fix the code style issue about open brace {
Peng Li [Tue, 15 Jun 2021 02:43:44 +0000 (10:43 +0800)]
net: z85230: fix the code style issue about open brace {

This patch fixes the code style issue according to checkpatch.pl error:
"ERROR: that open brace { should be on the previous line".

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: add some required spaces
Peng Li [Tue, 15 Jun 2021 02:43:43 +0000 (10:43 +0800)]
net: z85230: add some required spaces

Add space required before the open parenthesis '(' and '{'.
Add space required after that close brace '}' and ','
Add spaces required around that '=' , '&', '*', '|', '+', '/' and '-'.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: remove trailing whitespaces
Peng Li [Tue, 15 Jun 2021 02:43:42 +0000 (10:43 +0800)]
net: z85230: remove trailing whitespaces

This patch removes trailing whitespaces.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: fix the code style issue about "if..else.."
Peng Li [Tue, 15 Jun 2021 02:43:41 +0000 (10:43 +0800)]
net: z85230: fix the code style issue about "if..else.."

According to the chackpatch.pl, else should follow close brace '}',
braces {} should be used on all arms of this statement.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: fix the comments style issue
Peng Li [Tue, 15 Jun 2021 02:43:40 +0000 (10:43 +0800)]
net: z85230: fix the comments style issue

Networking block comments don't use an empty /* line,
use /* Comment...

Block comments use * on subsequent lines.
Block comments use a trailing */ on a separate line.

This patch fixes the comments style issues.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: replace comparison to NULL with "!skb"
Peng Li [Tue, 15 Jun 2021 02:43:39 +0000 (10:43 +0800)]
net: z85230: replace comparison to NULL with "!skb"

According to the chackpatch.pl, comparison to NULL could
be written "!skb".

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: fix the code style issue about EXPORT_SYMBOL(foo)
Peng Li [Tue, 15 Jun 2021 02:43:38 +0000 (10:43 +0800)]
net: z85230: fix the code style issue about EXPORT_SYMBOL(foo)

According to the chackpatch.pl,
EXPORT_SYMBOL(foo); should immediately follow its function/variable.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: add blank line after declarations
Peng Li [Tue, 15 Jun 2021 02:43:37 +0000 (10:43 +0800)]
net: z85230: add blank line after declarations

This patch fixes the checkpatch error about missing a blank line
after declarations.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: z85230: remove redundant blank lines
Peng Li [Tue, 15 Jun 2021 02:43:36 +0000 (10:43 +0800)]
net: z85230: remove redundant blank lines

This patch removes some redundant blank lines.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet/sched: cls_flower: Remove match on n_proto
Boris Sukholitko [Mon, 14 Jun 2021 11:13:22 +0000 (14:13 +0300)]
net/sched: cls_flower: Remove match on n_proto

The following flower filters fail to match packets:

tc filter add dev eth0 ingress protocol 0x8864 flower \
action simple sdata hi64
tc filter add dev eth0 ingress protocol 802.1q flower \
vlan_ethtype 0x8864 action simple sdata "hi vlan"

The protocol 0x8864 (ETH_P_PPP_SES) is a tunnel protocol. As such, it is
being dissected by __skb_flow_dissect and it's internal protocol is
being set as key->basic.n_proto. IOW, the existence of ETH_P_PPP_SES
tunnel is transparent to the callers of __skb_flow_dissect.

OTOH, in the filters above, cls_flower configures its key->basic.n_proto
to the ETH_P_PPP_SES value configured by the user. Matching on this key
fails because of __skb_flow_dissect "transparency" mentioned above.

In the following, I would argue that the problem lies with cls_flower,
unnessary attempting key->basic.n_proto match.

There are 3 close places in fl_set_key in cls_flower setting up
mask->basic.n_proto. They are (in reverse order of appearance in the
code) due to:

(a) No vlan is given: use TCA_FLOWER_KEY_ETH_TYPE parameter
(b) One vlan tag is given: use TCA_FLOWER_KEY_VLAN_ETH_TYPE
(c) Two vlans are given: use TCA_FLOWER_KEY_CVLAN_ETH_TYPE

The match in case (a) is unneeded because flower has no its own
eth_type parameter. It was removed by Jamal Hadi Salim in commit
488b41d020fb06428b90289f70a41210718f52b7 in iproute2. For
TCA_FLOWER_KEY_ETH_TYPE the userspace uses the generic tc filter
protocol field. Therefore the match for the case (a) is done by tc
itself.

The matches in cases (b), (c) are unneeded because the protocol will
appear in and will be matched by flow_dissector_key_vlan.vlan_tpid.
Therefore in the best case, key->basic.n_proto will try to repeat vlan
key match again.

The below patch removes mask->basic.n_proto setting and resets it to 0
in case (c).

Signed-off-by: Boris Sukholitko <boris.sukholitko@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agostmmac: align RX buffers
Matteo Croce [Mon, 14 Jun 2021 02:25:04 +0000 (04:25 +0200)]
stmmac: align RX buffers

On RX an SKB is allocated and the received buffer is copied into it.
But on some architectures, the memcpy() needs the source and destination
buffers to have the same alignment to be efficient.

This is not our case, because SKB data pointer is misaligned by two bytes
to compensate the ethernet header.

Align the RX buffer the same way as the SKB one, so the copy is faster.
An iperf3 RX test gives a decent improvement on a RISC-V machine:

before:
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec   733 MBytes   615 Mbits/sec   88             sender
[  5]   0.00-10.01  sec   730 MBytes   612 Mbits/sec                  receiver

after:
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec  1.10 GBytes   942 Mbits/sec    0             sender
[  5]   0.00-10.00  sec  1.09 GBytes   940 Mbits/sec                  receiver

And the memcpy() overhead during the RX drops dramatically.

before:
Overhead  Shared O  Symbol
  43.35%  [kernel]  [k] memcpy
  33.77%  [kernel]  [k] __asm_copy_to_user
   3.64%  [kernel]  [k] sifive_l2_flush64_range

after:
Overhead  Shared O  Symbol
  45.40%  [kernel]  [k] __asm_copy_to_user
  28.09%  [kernel]  [k] memcpy
   4.27%  [kernel]  [k] sifive_l2_flush64_range

Signed-off-by: Matteo Croce <mcroce@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet/mlx5: Round-Robin EQs over IRQs
Shay Drory [Tue, 23 Feb 2021 09:57:32 +0000 (11:57 +0200)]
net/mlx5: Round-Robin EQs over IRQs

Whenever users provided affinity for an EQ creation request, map the
EQ to a matching IRQ.
Matching IRQ=IRQ with the same affinity and type (completion/control) of
the EQ created.

This mapping is being done in agressive dedicated IRQ allocation scheme,
which described bellow.

First, we check whether there is a matching IRQ that his min threshold
is not exhausted.
   - min_eqs_threshold = 3 for control EQ.
   - min_eqs_threshold = 1 for completion EQ.
In case no matching IRQ was found, try to request a new IRQ.
In case we can't request a new IRQ, reuse least-used matching IRQ.

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: Separate between public and private API of sf.h
Shay Drory [Tue, 11 May 2021 15:48:30 +0000 (18:48 +0300)]
net/mlx5: Separate between public and private API of sf.h

Move mlx5_sf_max_functions() and friends from the privete sf/sf.h
to the public lib/sf.h. This is done in order to have one direction
include paths.

Signed-off-by: Shay Drory <shayd@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: Enlarge interrupt field in CREATE_EQ
Shay Drory [Mon, 10 May 2021 06:10:43 +0000 (09:10 +0300)]
net/mlx5: Enlarge interrupt field in CREATE_EQ

FW is now supporting more than 256 MSI-X per PF (up to 2K).
Hence, enlarge interrupt field in CREATE_EQ to make use of the new
MSI-X's.

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: Allocating a pool of MSI-X vectors for SFs
Shay Drory [Tue, 23 Feb 2021 09:48:17 +0000 (11:48 +0200)]
net/mlx5: Allocating a pool of MSI-X vectors for SFs

SFs (Sub Functions) currently use IRQs from the global IRQ table their
parent Physical Function have. In order to better scale, we need to
allocate more IRQs and share them between different SFs.

Driver will maintain 3 separated irq pools:
1. A pool that serve the PF consumer (PF's netdev, rdma stacks), similar
to what the driver had before this patch. i.e, this pool will share irqs
between rdma and netev, and will keep the irq indexes and allocation
order. The last is important for PF netdev rmap (aRFS).

2. A pool of control IRQs for SFs. The size of this pool is the number
of SFs that can be created divided by SFS_PER_IRQ. This pool will serve
the control path EQs of the SFs.

3. A pool of completion data path IRQs for SFs transport queues. The
size of this pool is:
num_irqs_allocated - pf_pool_size - sf_ctrl_pool_size.
This pool will served netdev and rdma stacks. Moreover, rmap is not
supported on SFs.

Sharing methodology of the SFs pools is explained in the next patch.

Important note: rmap is not supported on SFs because rmap mapping cannot
function correctly for IRQs that are shared for different core/netdev RX
rings.

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: Change IRQ storage logic from static to dynamic
Shay Drory [Tue, 23 Feb 2021 09:38:52 +0000 (11:38 +0200)]
net/mlx5: Change IRQ storage logic from static to dynamic

Store newly created IRQs in the xarray DB instead of a static array,
so we will be able to store only IRQs which are being used.

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: Moving rmap logic to EQs
Shay Drory [Tue, 23 Feb 2021 09:24:47 +0000 (11:24 +0200)]
net/mlx5: Moving rmap logic to EQs

IRQs are being simplified in order to ease their sharing and any feature
specific object will be moved to upper layer.
Hence we move rmap object into eq_table.

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>