Function netdev_master_upper_dev_get() generates warning if caller doesn't
hold rtnl lock. Modify rules update path to use rcu version of that
function.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Jianbo Liu <jianbol@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
esw->state_lock is already used to protect vlan vport configuration change.
However, all preparation and correctness checks, and code that sets vport
data are not protected by this lock and assume external synchronization by
rtnl lock. In order to remove dependency on rtnl lock, extend
esw->state_lock protection to whole eswitch vlan add/del functions.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Jianbo Liu <jianbol@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Eswitch implements its own locking by means of state_lock mutex and
multiple fine-grained lock in containing data structures, and is supposed
to not rely on rtnl lock. However, eswitch offloads num_flows type is a
regular long long integer and cannot be modified concurrently. This is an
implicit assumptions that mlx5 tc is serialized (by rtnl lock or any other
means). In order to remove implicit dependency on rtnl lock, change
num_flows type to atomic64 to allow concurrent modifications.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Jianbo Liu <jianbol@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
In order to remove dependency on rtnl lock for protecting unready_flows
list when reoffloading unready flows on workqueue, extend representor
uplink private structure with dedicated 'unready_flows_lock' mutex. Take
the lock in all users of unready_flows list before accessing it. Implement
helper functions to add and delete unready flow.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Jianbo Liu <jianbol@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
In order to remove dependency on rtnl lock, access to tc flows hashtable
must be explicitly protected from concurrent flows removal.
Extend tc flow structure with rcu to allow concurrent parallel access. Use
rcu read lock to safely lookup flow in tc flows hash table, and take
reference to it. Use rcu free for flow deletion to accommodate concurrent
stats requests.
Add new DELETED flow flag. Imlement new flow_flag_test_and_set() helper
that is used to set a flag and return its previous value. Use it to
atomically set the flag in mlx5e_delete_flower() to guarantee that flow can
only be deleted once, even when same flow is deleted concurrently by
multiple tasks.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Jianbo Liu <jianbol@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
To remove dependency on rtnl lock and allow concurrent modification of
'flags' field of tc flow structure, change flow flag type to unsigned long
and use atomic bit ops for reading and changing the flags. Implement
auxiliary functions for setting, resetting and getting specific flag, and
for checking most often used flag values.
Always set flags with smp_mb__before_atomic() to ensure that all
mlx5e_tc_flow are updated before concurrent readers can read new flags
value. Rearrange all code paths to actually set flow->rule[] pointers
before setting the OFFLOADED flag. On read side, use smp_mb__after_atomic()
when accessing flags to ensure that offload-related flow fields are only
read after the flags.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Jianbo Liu <jianbol@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
With new classifier type that doesn't require rtnl lock, following
invariant holds:
- Filter with specified cookie created only once.
- Filter with specified cookie deleted only once.
- Stats updates can be performed in parallel to each other.
Extend tc flow with rcu and reference counter. To protect from concurrent
delete, get reference to tc flow when:
- Reading flow stats.
- Accessing flow in neigh update handler.
- Accessing flow in neigh update used value handler.
Only free flow when reference counter reached zero. Modify flow cleanup to
account for flows that could be not fully initialized by checking if flow
is actually in the list of corresponding mod_hdr, hairpin and encap
entries. Don't cleanup flow directly in case of error to allow concurrent
neigh update (neigh update will be modified to always take reference to
flow when using it).
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Jianbo Liu <jianbol@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
The helper function has "if" branches that do the same. Merge them to
simplify the code.
Signed-off-by: Eli Britstein <elibr@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
When call flow_block_cb_is_busy. The indr_priv is guaranteed to
NULL ptr. So there is no need to call flow_bock_cb_is_busy.
Fixes: 0d4fd02e71 ("net: flow_offload: add flow_block_cb_is_busy() and use it")
Signed-off-by: wenxu <wenxu@ucloud.cn>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Don't choose who implements the rxnfc "get/set" callbacks according to
CONFIG_MLX5_EN_RXNFC, instead have the callbacks always available and
delegate to a function of a different driver module when needed
(en_fs_ethtool.c), have stubs in en/fs.h to fallback to when
en_fs_ethtool.c is compiled out, to avoid complications and ifdefs in
en_main.c.
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
When disabling CQE compression in favor of time-stamping, don't show a
warning when CQE compression is already disabled.
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
When user enables LRO via ethtool and if the RQ mode is legacy,
mlx5e_fix_features drops the request without any explanation.
Add netdev_warn to cover this case.
Fixes: 6c3a823e1e ("net/mlx5e: RX, Remove HW LRO support in legacy RQ")
Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Define the 57508, 57504, and 57502 chip IDs that are all part of the
BNXT_CHIP_P5 family of chips.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the new TPA feature in the 57500 chips, we need to discover the
feature first before setting up the netdev features. Refactor the
the firmware probe and init logic more cleanly into 2 functions and
and make these calls before setting up the netdev features.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Support the new expanded TPA v2 counters on 57500 B0 chips for
ethtool -S.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The new TPA implemantation has additional TPA counters that extend the
per-ring statistics block. Allocate the proper size accordingly.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The current code assumes that the per ring statistics counters are
fixed. In newer chips that support a newer version of TPA, the
TPA counters are also changed. Refactor the code by defining these
counter names in arrays so that it is easy to add a new array for
a new set of counters supported by the newer chips.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a more optimized hardware GRO function to setup the SKB on 57500
chips. Some workaround code is no longer needed on 57500 chips and
the pseudo checksum is also calculated in hardware, so no need to
do the software pseudo checksum in the driver.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The new TPA feature on 57500 supports a larger number of concurrent TPAs
(up to 1024) divided among the functions. We need to add some logic to
map the hardware TPA ID to a software index that keeps track of each TPA
in progress. A 1:1 direct mapping without translation would be too
wasteful as we would have to allocate 1024 TPA structures for each RX
ring on each PCI function.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With all the previous refactoring, the TPA fast path can now be
modified slightly to support TPA on the new chips. The main
difference is that the agg completions are retrieved differently using
the bnxt_get_tpa_agg_p5() function on the new chips.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
On 57500 chips, hardware GRO mode cannot be determined from the TPA
end, so we need to check bp->flags to determine if we are in hardware
GRO mode or not. Modify bnxt_set_features so that the TPA flags
in bp->flags don't change until the device is closed. This will ensure
that the fast path can safely rely on bp->flags to determine the
TPA mode.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The 2 GRO functions to set up the hardware GRO SKB fields for 2
different hardware chips have practically identical logic for
tunneled packets. Refactor the logic into a separate bnxt_gro_tunnel()
function that can be used by both functions.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
On the new 57500 chips, these new RX_AGG completions are not coalesced
at the TPA_END completion. Handle these by storing them in the
array in the bnxt_tpa_info struct, as they are seen when processing
the CMPL ring.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add an aggregation array to bnxt_tpa_info struct to keep track of the
aggregation completions. The aggregation completions are not
completed at the TPA_END completion on 57500 chips so we need to
keep track of them. The array is only allocated on the new chips
when required. An agg_count field is also added to keep track of the
number of these completions.
The maximum concurrent TPA is now discovered from firmware instead of
the hardcoded 64. Add a new bp->max_tpa to keep track of maximum
configured TPA.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Refactor the TPA logic slightly, so that the code can be more easily
extended to support TPA on the new 57500 chips. In particular, the
logic to get the next aggregation completion is refactored into a
new function bnxt_get_agg() so that this operation is made more
generalized. This operation will be different on the new chip in TPA
mode. The logic to recycle the aggregation buffers has a new start
index parameter added for the same purpose.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The new chips have a slightly modified TPA interface for LRO/GRO_HW.
Modify the TPA structures so that the same structures can also be
used on the new chips.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Among the changes are new CoS discard counters and new ctx_hw_stats_ext
struct for the latest 5750X B0 chips.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We have an ERPS (Ethernet Ring Protection Switching) setup involving
mv88e6250 switches which we're in the process of switching to a BSP
based on the mainline driver. Breaking any link in the ring works as
expected, with the ring reconfiguring itself quickly and traffic
continuing with almost no noticable drops. However, when plugging back
the cable, we see 5+ second stalls.
This has been tracked down to the userspace application in charge of
the protocol missing a few CCM messages on the good link (the one that
was not unplugged), causing it to broadcast a "signal fail". That
message eventually reaches its link partner, which responds by
blocking the port. Meanwhile, the first node has continued to block
the port with the just plugged-in cable, breaking the network. And the
reason for those missing CCM messages has in turn been tracked down to
the VTU apparently being too busy servicing load/purge operations that
the normal lookups are delayed.
Initial state, the link between C and D is blocked in software.
_____________________
/ \
| |
A ----- B ----- C *---- D
Unplug the cable between C and D.
_____________________
/ \
| |
A ----- B ----- C * * D
Reestablish the link between C and D.
_____________________
/ \
| |
A ----- B ----- C *---- D
Somehow, enough VTU/ATU operations happen inside C that prevents
the application from receving the CCM messages from B in a timely
manner, so a Signal Fail message is sent by C. When B receives
that, it responds by blocking its port.
_____________________
/ \
| |
A ----- B *---* C *---- D
Very shortly after this, the signal fail condition clears on the
BC link (some CCM messages finally make it through), so C
unblocks the port. However, a guard timer inside B prevents it
from removing the blocking before 5 seconds have elapsed.
It is not unlikely that our userspace ERPS implementation could be
smarter and/or is simply buggy. However, this patch fixes the symptoms
we see, and is a small optimization that should not break anything
(knock wood). The idea is simply to avoid doing an VTU load of an
entry identical to the one already present. To do that, we need to
know whether mv88e6xxx_vtu_get() actually found an existing entry, or
has just prepared a struct mv88e6xxx_vtu_entry for us to load. To that
end, let vlan->valid be an output parameter. The other two callers of
mv88e6xxx_vtu_get() are not affected by this patch since they pass
new=false.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
There was a previous attempt to use xmit_more, but the change had to be
reverted because under load sometimes a transmit timeout occurred [0].
Maybe this was caused by a missing memory barrier, the new attempt
keeps the memory barrier before the call to netif_stop_queue like it
is used by the driver as of today. The new attempt also changes the
order of some calls as suggested by Eric.
[0] https://lkml.org/lkml/2019/2/10/39
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
use resource_size to calcuate ioremap size and make
the code simpler.
Signed-off-by: Ding Xiang <dingxiang@cmss.chinamobile.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
dev_info() is more appropriate for printing messages when driver
initialization done, so switch to dev_info().
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Hardware supports 0.5K, 1K, 2K, 4K RX buffer size, the
RX buffer can not be reused because the hns3_page_order
return 0 when page size and RX buffer size are both 4096.
So this patch changes the hns3_page_order to return 1 when
RX buffer is greater than half of the page size and page size
is less the 8192, and dev_alloc_pages has already been used
to allocate the compound page for RX buffer.
This patch also changes hnae3_* to hns3_* for page order
and RX buffer size calculation because they are used in
hns3 module.
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Reviewed-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The misc interrupt is used to schedule the reset and mailbox
subtask, and service_task delayed_work is used to do periodic
management work each second.
This patch sets the above three subtask's affinity using the
misc interrupt' affinity.
Also this patch setups a affinity notify for misc interrupt to
allow user to change the above three subtask's affinity.
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use delayed work instead of using timers to trigger the
hclge_serive.
Simplify the code with one less middle function and in order
to support misc irq affinity.
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Reviewed-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some times just see the eth interface have been down/up via
dmesg, but can not know why the eth down. So adds some debug
messages to identify the cause for this.
Signed-off-by: Yonglong Liu <liuyonglong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch modifies firmware version display format in
hclge(vf)_cmd_init() and hns3_get_drvinfo(). Also, adds
some optimizations for firmware version display format.
Signed-off-by: Yufeng Mo <moyufeng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When allocating memory, the GFP_KERNEL cannot be used during the
spin_lock period. This is because it may cause scheduling when holding
spin_lock. This patch changes GFP flag to GFP_ATOMIC in this case.
Fixes: dd74f815dd ("net: hns3: Add support for rule add/delete for flow director")
Signed-off-by: Yufeng Mo <moyufeng@huawei.com>
Signed-off-by: lipeng 00277521 <lipeng321@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, hclge_reset_err_handle() will assert a global reset
when the failing count is smaller than MAX_RESET_FAIL_CNT, which
will affect other running functions.
So this patch removes this upgrading, and uses re-scheduling reset
task to do it.
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Reviewed-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For some cases, ops->get_reset_level may not be implemented, so we
should check whether it is NULL before calling get_reset_level.
Signed-off-by: Guangbin Huang <huangguangbin@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
hns3_set_channels() should check the resetting status firstly,
since the device will reinitialize when resetting. If the
reset has not completed, the hns3_set_channels() may access
invalid memory.
Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Matching on reserved TCP flags bits is only supported using custom
parser. Since the usecase for that is not known now, just forbid to
offload rules that match on these bits.
Reported-by: Alex Kushnarov <alexanderk@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some matches and actions are not supported on egress. Track such rules
and forbid a bind of block which contains them to egress.
With this patch, the kernel tells the user he cannot do that:
$ tc qdisc add dev ens16np1 ingress_block 22 clsact
$ tc filter add block 22 protocol 802.1q pref 2 handle 101 flower vlan_id 100 skip_sw action pass
$ tc qdisc add dev ens16np2 egress_block 22 clsact
Error: mlxsw_spectrum: Block cannot be bound to egress because it contains unsupported rules.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Spectrum ASIC does not support redirection on egress, so refuse to
insert such flows:
$ tc qdisc add dev ens16np1 clsact
$ tc filter add dev ens16np1 egress protocol all pref 1 handle 101 flower skip_sw action mirred egress redirect dev ens16np2
Error: mlxsw_spectrum: Redirect action is not supported on egress.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Enable HW csum and TSO per default except on known buggy chip versions.
Realtek confirmed that RTL8168evl has a HW issue with TSO.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The loop in r8169_csum_workaround is called only if in
msdn_giant_send_check a copy of the skb header needs to be made and
we don't have enough memory. Let's simply drop the packet in that case
so that we can remove r8169_csum_workaround.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement callback ndo_features_check and move all feature checks there.
This will allow us to get rid of r8169_csum_workaround() completely in
a subsequent step. Like in the vendor driver disable HW csum for short
packets on RTL8168b.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Set GSO max size and max segment number as in the vendor driver.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Align setting PME with the vendor driver. PMEnable is writable on
RTL8169 only, on later chip versions it's read-only. PME_SIGNAL is
used on chip versions from RTL8168evl with the exception of the
RTL8168f family.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ConnectX-3 Pro can offload transmission of VLAN packets with VXLAN inside:
enable tunnel offloads in dev->vlan_features, like it's done with other
NIC drivers (e.g. be2net and ixgbe).
It's no more necessary to change dev->hw_enc_features when VXLAN are added
or removed, since .ndo_features_check() already checks for VXLAN packet
where the UDP destination port matches the configured value. Just set
dev->hw_enc_features when the NIC is initialized, so that overlying VLAN
can correctly inherit the tunnel offload capabilities.
Changes since v1:
- avoid flipping hw_enc_features, instead of calling netdev notifiers,
thanks to Saeed Mahameed
- squash two patches into a single one
CC: Paolo Abeni <pabeni@redhat.com>
CC: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Reviewed-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
