There are two reasons for this.
First, the xmit_more flag conceptually doesn't fit into the skb, as
xmit_more is not a property related to the skb.
Its only a hint to the driver that the stack is about to transmit another
packet immediately.
Second, it was only done this way to not have to pass another argument
to ndo_start_xmit().
We can place xmit_more in the softnet data, next to the device recursion.
The recursion counter is already written to on each transmit. The "more"
indicator is placed right next to it.
Drivers can use the netdev_xmit_more() helper instead of skb->xmit_more
to check the "more packets coming" hint.
skb->xmit_more is retained (but always 0) to not cause build breakage.
This change takes care of the simple s/skb->xmit_more/netdev_xmit_more()/
conversions. Remaining drivers are converted in the next patches.
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that the NSP provides the ability to read from the SFF modules'
EEPROM, we can use this interface to implement the ethtool callback.
If the NSP only provides partial data, we log the event from within
the driver but pass a success code to ethtool to prevent it from
discarding the partial data.
Signed-off-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The NSP now provides the ability to read from the SFF module EEPROM.
Note that even if an error occurs, the NSP may still provide some of the
data.
Signed-off-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the introduction of flow_action_for_each pedit actions are no
longer grouped together, instead pedit actions are broken out per
32 byte word. This results in an inefficient use of the action list
that is pushed to hardware where each 32 byte word becomes its own
action. Therefore we combine groups of 32 byte word before sending
the action list to hardware.
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We no longer set CFI when pushing vlan tags, therefore we remove
the CFI bit from push vlan.
Fixes: 1a1e586f54 ("nfp: add basic action capabilities to flower offloads")
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Signed-off-by: Louis Peens <louis.peens@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace vlan CFI bit with a vlan present bit that indicates the
presence of a vlan tag. Previously the driver incorrectly assumed
that an vlan id of 0 is not matchable, therefore we indicate vlan
presence with a vlan present bit.
Fixes: 5571e8c9f2 ("nfp: extend flower matching capabilities")
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Signed-off-by: Louis Peens <louis.peens@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Switch to readl/writel() APIs, because this is recommended
API and the MDIO block is reused on Keystone 2 SoCs
where LE/BE modes are supported.
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
One of the more common cases of allocation size calculations is finding
the size of a structure that has a zero-sized array at the end, along
with memory for some number of elements for that array. For example:
struct foo {
int stuff;
struct boo entry[];
};
size = sizeof(struct foo) + count * sizeof(struct boo);
instance = kvzalloc(size, GFP_KERNEL);
Instead of leaving these open-coded and prone to type mistakes, we can
now use the new struct_size() helper:
instance = kvzalloc(struct_size(instance, entry, count), GFP_KERNEL);
Notice that, in this case, variable size is not necessary, hence
it is removed.
This code was detected with the help of Coccinelle.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The use of zero-sized array causes undefined behaviour when it is not
the last member in a structure. As it happens to be in this case.
Also, the current code makes use of a language extension to the C90
standard, but the preferred mechanism to declare variable-length
types such as this one is a flexible array member, introduced in
C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last. Which is beneficial
to cultivate a high-quality code.
Fixes: e48f129c2f ("[SCSI] cxgb3i: convert cdev->l2opt to use rcu to prevent NULL dereference")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit f3fef2b6e1 ("i40e: Remove umem from VSI") a regression was
introduced; When the VSI was reset, the setup code would try to enable
AF_XDP ZC unconditionally (as long as there was a umem placed in the
netdev._rx struct). Here, we add a bitmap to the VSI that tracks if a
certain queue pair has been "zero-copy enabled" via the ndo_bpf. The
bitmap is used in i40e_xsk_umem, and enables zero-copy if and only if
XDP is enabled, the corresponding qid in the bitmap is set and the
umem is non-NULL.
Fixes: f3fef2b6e1 ("i40e: Remove umem from VSI")
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
The i40e_xsk_umem function was explicitly inlined in i40e.h. There is
no reason for that, so move it to i40e_main.c instead.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This is log is harmful as it can trigger multiple times per packet. Delete
it.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Packets without the last descriptor set should be dropped early. If we
receive a frame larger than the DMA buffer, the HW will continue using the
next descriptor. Driver mistakes these as individual frames, and sometimes
a truncated frame (without the LD set) may look like a valid packet.
This fixes a strange issue where the system replies to 4098-byte ping
although the MTU/DMA buffer size is set to 4096, and yet at the same
time it's logging an oversized packet.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If we have error bits set, the discard_frame status will get overwritten
by checksum bit checks, which might set the status back to good one.
Fix by checking the COE status only if the frame is good.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, if we drop a packet, we exit from NAPI loop before the budget
is consumed. In some situations this will make the RX processing stall
e.g. when flood pinging the system with oversized packets, as the
errorneous packets are not dropped efficiently.
If we drop a packet, we should just continue to the next one as long as
the budget allows.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We always program the maximum DMA buffer size into the receive descriptor,
although the allocated size may be less. E.g. with the default MTU size
we allocate only 1536 bytes. If somebody sends us a bigger frame, then
memory may get corrupted.
Fix by using exact buffer sizes.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use netif_receive_skb_list() instead of napi_gro_receive() to benefit
from batched skb processing.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The name of the trace is no longer correct, since there is no disable of
rehash done. So name it "rehash_rollback_failed".
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>
Currently if a rollback ends with error, the vregion is in a zombie
state until end of the existence. Instead of that, rather try to
continue where rollback ended later on (after rehash interval).
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>
Move the call of rehash_dis trace and the error message to
vregion_migrate() next to the failed_rollback flag set.
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>
The flag is set by the caller mlxsw_sp_acl_tcam_vregion_migrate() anyway,
so don't set it here.
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>
It was reported that re-introducing ASPM, in combination with RX
interrupt coalescing, results in significantly increased packet
latency, see [0]. Disabling ASPM or RX interrupt coalescing fixes
the issue. Therefore change the driver's default to disable RX
interrupt coalescing. Users still have the option to enable RX
coalescing via ethtool.
[0] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=925496
Fixes: a99790bf5c ("r8169: Reinstate ASPM Support")
Reported-by: Mike Crowe <mac@mcrowe.com>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
-----BEGIN PGP SIGNATURE-----
iQEcBAABAgAGBQJcnnD7AAoJEEg/ir3gV/o+nksIAImv3hLoPXJoS/hnHQJV4FtB
U6euKao9PlKzMGCb7mii3WtBcn9ULwxNmWTA6oiJhZl2gp1gS/efAK5RFAis78pe
FKCuFU07antVcJQZ118L2gKSPDHwxuEOHm4d0IqlxzXr6gzldCZD+jSekD+nB59g
7Ehex+wi5ChPz5JqNA9UM5VQ9lO9Mf4yVrXKaZG3wAnT9Y0w2qwvqeNx5xVbXao9
3ugObrK24WVnajOK8Ub1MwREgveTuzccUy5GB4gc6DIBI8oSwKh+3DdqGqwiCM0z
/l5NmxK1ma+YpoKs1p6G7ipyTRielJD4oMbJeoilHfxsFyA1+AOIzdHjBw4Cd/4=
=SQ1L
-----END PGP SIGNATURE-----
Merge tag 'mlx5-fixes-2019-03-29' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux
Saeed Mahameed says:
====================
Mellanox, mlx5 fixes 2019-03-29
This series introduces some fixes to mlx5 driver.
Please pull and let me know if there is any problem.
For -stable v4.11
('net/mlx5: Decrease default mr cache size')
For -stable v4.12
('net/mlx5e: Add a lock on tir list')
For -stable v4.13
('net/mlx5e: Fix error handling when refreshing TIRs')
For -stable v4.18
('net/mlx5e: Update xon formula')
For -stable v4.19
('net: mlx5: Add a missing check on idr_find, free buf')
('net/mlx5e: Update xoff formula')
net-next merge Note:
When merged with net-next the following simple conflict will appear,
drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
++<<<<<<< HEAD (net)
+ * max_mtu: netdev's max_mtu
++=======
+ * @mtu: device's MTU
++>>>>>>> net-next
To resolve: just replace the line in net-next
* @mtu: device's MTU
to
* @max_mtu: netdev's max_mtu
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch advertises Forward Error Correction in ethtool
Signed-off-by: Casey Leedom <leedom@chelsio.com>
Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Change t4fw_version.h to update latest firmware version
number to 1.23.3.0.
Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Change t4fw_version.h to update latest firmware version
number to 1.23.3.0.
Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The driver allocates an encap context based on the tunnel properties,
and reuse that context for all flows using the same tunnel properties.
Commit df2ef3bff1 ("net/mlx5e: Add GRE protocol offloading")
introduced another tunnel protocol other than the single VXLAN
previously supported. A flow that uses a tunnel with the same tunnel
properties but with a different tunnel type (GRE vs VXLAN for example)
would mistakenly reuse the previous alocated context, causing the
traffic to be sent with the wrong encapsulation. Fix that by
considering the tunnel type for encap contexts.
Fixes: df2ef3bff1 ("net/mlx5e: Add GRE protocol offloading")
Signed-off-by: Eli Britstein <elibr@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Set xon = xoff - netdev's max_mtu.
netdev's max_mtu will give enough time for the pause frame to
arrive at the sender.
Fixes: 0696d60853 ("net/mlx5e: Receive buffer configuration")
Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Make sure the struct mlx5_flow_destination is zero before
filling in the field.
Fixes: 8da202b249 ("net/mlx5: E-Switch, Add support for VEPA in legacy mode.")
Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Reviewed-by: Daniel Jurgens <danielj@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Traditionally, the PF (Physical Function) which resides on vport 0 was
the E-switch manager. Since the ECPF (Embedded CPU Physical Function),
which resides on vport 0xfffe, was introduced as the E-Switch manager,
the assumption that the E-switch manager is on vport 0 is incorrect.
Since the eswitch code already uses the actual vport value, all we
need is to always set other_vport=1.
Signed-off-by: Omri Kahalon <omrik@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
The esw offloads structures share a union with the legacy mode structs.
Reset the offloads struct to zero in init to protect from null
assumptions made by the legacy mode code.
Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
The capacity of FDB offloading and NIC offloading table are
different, and when allocating the pedit actions, we should
use the correct namespace type.
Fixes: c500c86b0c ("net/mlx5e: support for two independent packet edit actions")
Cc: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Acked-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
The esw fdb table has a union of legacy and offloads members.
So if we were in a certain esw mode we could set some memebers and not
set null which is fine as on destroy path and don't care.
But then moving from legacy to switchdev a second time, the cleanup flow
of legacy mode checks if a struct member was in use if it's not null so
we need to make sure to reset the code to null when we init legacy mode.
Fixes: 8da202b249 ("net/mlx5: E-Switch, Add support for VEPA in legacy mode.")
Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Huy Nguyen <huyn@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Allow configuration of legacy link-modes even when extended link-modes
are supported. This requires reading of legacy advertisement even when
extended link-modes are supported. Since legacy and extended
advertisement are mutually excluded, wait for empty reply from extended
advertisement before reading legacy advertisement.
Fixes: 6a89737241 ("net/mlx5: ethtool, Add ethtool support for 50Gbps per lane link modes")
Signed-off-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Ethtool option set_link_ksettings allows setting of legacy link-modes
or extended link-modes. Refine the decision of which type of link-modes
is set.
Fixes: 6a89737241 ("net/mlx5: ethtool, Add ethtool support for 50Gbps per lane link modes")
Signed-off-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Refresh tirs is looping over a global list of tirs while netdevs are
adding and removing tirs from that list. That is why a lock is
required.
Fixes: 724b2aa151 ("net/mlx5e: TIRs management refactoring")
Signed-off-by: Yuval Avnery <yuvalav@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
idr_find() can return a NULL value to 'flow' which is used without a
check. The patch adds a check to avoid potential NULL pointer dereference.
In case of mlx5_fpga_sbu_conn_sendmsg() failure, free buf allocated
using kzalloc.
Fixes: ab412e1dd7 ("net/mlx5: Accel, add TLS rx offload routines")
Signed-off-by: Aditya Pakki <pakki001@umn.edu>
Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
For some protocols we are not allowing IP header rewrite offload, since
the HW is not capable to properly adjust the l4 checksum. However, TTL
& HOPLIMIT modification can be done for all IP protocols, because they
are not part of the pseudo header taken into account for checksum.
Fixes: 7386788175 ("drivers: net: use flow action infrastructure")
Signed-off-by: Dmytro Linkin <dmitrolin@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Previously, a false positive would be caught if the TIRs list is
empty, since the err value was initialized to -ENOMEM, and was only
updated if a TIR is refreshed. This is resolved by initializing the
err value to zero.
Fixes: b676f65389 ("net/mlx5e: Refactor refresh TIRs")
Signed-off-by: Gavi Teitz <gavi@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Delete initialization of high order entries in mr cache to decrease initial
memory footprint. When required, the administrator can populate the
entries with memory keys via the /sys interface.
This approach is very helpful to significantly reduce the per HW function
memory footprint in virtualization environments such as SRIOV.
Fixes: 9603b61de1 ("mlx5: Move pci device handling from mlx5_ib to mlx5_core")
Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reported-by: Shalom Toledo <shalomt@mellanox.com>
Acked-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
The number of stubs is growing and has nothing to do with addrconf.
Move the definition of the stubs to a separate header file and update
users. In the move, drop the vxlan specific comment before ipv6_stub.
Code move only; no functional change intended.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Rename fib6_nh entries that will be moved to a fib_nh_common struct.
Specifically, the device, gateway, flags, and lwtstate are common
with all nexthop definitions. In some places new temporary variables
are declared or local variables renamed to maintain line lengths.
Rename only; no functional change intended.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Rename fib_nh entries that will be moved to a fib_nh_common struct.
Specifically, the device, oif, gateway, flags, scope, lwtstate,
nh_weight and nh_upper_bound are common with all nexthop definitions.
In the process shorten fib_nh_lwtstate to fib_nh_lws to avoid really
long lines.
Rename only; no functional change intended.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The gateway setting is not per fib6_info entry but per-fib6_nh. Add a new
fib_nh_has_gw flag to fib6_nh and convert references to RTF_GATEWAY to
the new flag. For IPv6 address the flag is cheaper than checking that
nh_gw is non-0 like IPv4 does.
While this increases fib6_nh by 8-bytes, the effective allocation size of
a fib6_info is unchanged. The 8 bytes is recovered later with a
fib_nh_common change.
Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
NFP reprs are software device on top of the PF's vNIC.
The comment above __dev_queue_xmit() sayeth:
When calling this method, interrupts MUST be enabled. This is because
the BH enable code must have IRQs enabled so that it will not deadlock.
For netconsole we can't guarantee IRQ state, let's just
disable netpoll on representors to be on the safe side.
When the initial implementation of NFP reprs was added by the
commit 5de73ee467 ("nfp: general representor implementation")
.ndo_poll_controller was required for netpoll to be enabled.
Fixes: ac3d9dd034 ("netpoll: make ndo_poll_controller() optional")
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
dev_queue_xmit() may return error codes as well as netdev_tx_t,
and it always consumes the skb. Make sure we always return a
correct netdev_tx_t value.
Fixes: eadfa4c3be ("nfp: add stats and xmit helpers for representors")
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some PHYs will use the 2500BaseX PHY_INTERFACE_MODE when being linked
with a partner using 2.5GBaseT.
Since we can't autonegotiate this speed between the MAC and the PHY, we
need to have the proper comphy support enabled, to make sure we can
safely advertise 2.5G and 1G in BaseT and be able to switch between both
corresponding PHY interface modes. This is now possible since comphy
support was added to this driver.
This commit adds the 2500BaseT mode to the list of supported modes when
using 2500BaseX, and was tested on a setup with an Armada385 and a
88E2010 PHY, both with and without the comphy node in the DT.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
