Skip to content

L
linux
Switch branch/tag
  1. 21 Oct, 2021 15 commits
    • David S. Miller's avatar
      Merge branch 'dsa_to_port-loops' · ce272973
      David S. Miller authored 
      Vladimir Oltean says:

====================
Remove the "dsa_to_port in a loop" antipattern

v1->v2: more patches
v2->v3: less patches

As opposed to previous series, I would now like to first refactor the
DSA core, since that sees fewer patches than drivers, and make the
helpers available. Since the refactoring is fairly noisy, I don't want
to force it on driver maintainers right away, patches can be submitted
independently.

The original cover letter is below:

The DSA core and drivers currently iterate too much through the port
list of a switch. For example, this snippet:

	for (port = 0; port < ds->num_ports; port++) {
		if (!dsa_is_cpu_port(ds, port))
			continue;

		ds->ops->change_tag_protocol(ds, port, tag_ops->proto);
	}

iterates through ds->num_ports once, and then calls dsa_is_cpu_port to
filter out the other types of ports. But that function has a hidden call
to dsa_to_port() in it, which contains:

	list_for_each_entry(dp, &dst->ports, list)
		if (dp->ds == ds && dp->index == p)
			return dp;

where the only thing we wanted to know in the first place was whether
dp->type == DSA_PORT_TYPE_CPU or not.

So it seems that the problem is that we are not iterating with the right
variable. We have an "int port" but in fact need a "struct dsa_port *dp".

This has started being an issue since this patch series:
https://patchwork.ozlabs.org/project/netdev/cover/20191020031941.3805884-1-vivien.didelot@gmail.com/

The currently proposed set of changes iterates like this:

	dsa_switch_for_each_cpu_port(cpu_dp, ds)
		err = ds->ops->change_tag_protocol(ds, cpu_dp->index,
						   tag_ops->proto);

which iterates directly over ds->dst->ports, which is a list of struct
dsa_port *dp. This makes it much easier and more efficient to check
dp->type.

As a nice side effect, with the proposed driver API, driver writers are
now encouraged to use more efficient patterns, and not only due to less
iterations through the port list. For example, something like this:

	for (port = 0; port < ds->num_ports; port++)
		do_something();

probably does not need to do_something() for the ports that are disabled
in the device tree. But adding extra code for that would look like this:

	for (port = 0; port < ds->num_ports; port++) {
		if (!dsa_is_unused_port(ds, port))
			continue;

		do_something();
	}

and therefore, it is understandable that some driver writers may decide
to not bother. This patch series introduces a "dsa_switch_for_each_available_port"
macro which comes at no extra cost in terms of lines of code / number of
braces to the driver writer, but it has the "dsa_is_unused_port" check
embedded within it.
====================
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      ce272973
    • Vladimir Oltean's avatar
      net: dsa: tag_8021q: make dsa_8021q_{rx,tx}_vid take dp as argument · 992e5cc7
      Vladimir Oltean authored 
      Pass a single argument to dsa_8021q_rx_vid and dsa_8021q_tx_vid that
contains the necessary information from the two arguments that are
currently provided: the switch and the port number.

Also rename those functions so that they have a dsa_port_* prefix, since
they operate on a struct dsa_port *.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      992e5cc7
    • Vladimir Oltean's avatar
      net: dsa: tag_sja1105: do not open-code dsa_switch_for_each_port · 5068887a
      Vladimir Oltean authored 
      Find the remaining iterators over dst->ports that only filter for the
ports belonging to a certain switch, and replace those with the
dsa_switch_for_each_port helper that we have now.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      5068887a
    • Vladimir Oltean's avatar
      net: dsa: convert cross-chip notifiers to iterate using dp · fac6abd5
      Vladimir Oltean authored 
      The majority of cross-chip switch notifiers need to filter in some way
over the type of ports: some install VLANs etc on all cascade ports.

The difference is that the matching function, which filters by port
type, is separate from the function where the iteration happens. So this
patch needs to refactor the matching functions' prototypes as well, to
take the dp as argument.

In a future patch/series, I might convert dsa_towards_port to return a
struct dsa_port *dp too, but at the moment it is a bit entangled with
dsa_routing_port which is also used by mv88e6xxx and they both return an
int port. So keep dsa_towards_port the way it is and convert it into a
dp using dsa_to_port.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      fac6abd5
    • Vladimir Oltean's avatar
      net: dsa: remove gratuitous use of dsa_is_{user,dsa,cpu}_port · 57d77986
      Vladimir Oltean authored 
      Find the occurrences of dsa_is_{user,dsa,cpu}_port where a struct
dsa_port *dp was already available in the function scope, and replace
them with the dsa_port_is_{user,dsa,cpu} equivalent function which uses
that dp directly and does not perform another hidden dsa_to_port().
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      57d77986
    • Vladimir Oltean's avatar
      net: dsa: do not open-code dsa_switch_for_each_port · 65c563a6
      Vladimir Oltean authored 
      Find the remaining iterators over dst->ports that only filter for the
ports belonging to a certain switch, and replace those with the
dsa_switch_for_each_port helper that we have now.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      65c563a6
    • Vladimir Oltean's avatar
      net: dsa: remove the "dsa_to_port in a loop" antipattern from the core · d0004a02
      Vladimir Oltean authored 
      Ever since Vivien's conversion of the ds->ports array into a dst->ports
list, and the introduction of dsa_to_port, iterations through the ports
of a switch became quadratic whenever dsa_to_port was needed.

dsa_to_port can either be called directly, or indirectly through the
dsa_is_{user,cpu,dsa,unused}_port helpers.

Use the newly introduced dsa_switch_for_each_port() iteration macro
that works with the iterator variable being a struct dsa_port *dp
directly, and not an int i. It is an expensive variable to go from i to
dp, but cheap to go from dp to i.

This macro iterates through the entire ds->dst->ports list and filters
by the ports belonging just to the switch provided as argument.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: default avatarFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      d0004a02
    • Vladimir Oltean's avatar
      net: dsa: introduce helpers for iterating through ports using dp · 82b31898
      Vladimir Oltean authored 
      Since the DSA conversion from the ds->ports array into the dst->ports
list, the DSA API has encouraged driver writers, as well as the core
itself, to write inefficient code.

Currently, code that wants to filter by a specific type of port when
iterating, like {!unused, user, cpu, dsa}, uses the dsa_is_*_port helper.
Under the hood, this uses dsa_to_port which iterates again through
dst->ports. But the driver iterates through the port list already, so
the complexity is quadratic for the typical case of a single-switch
tree.

This patch introduces some iteration helpers where the iterator is
already a struct dsa_port *dp, so that the other variant of the
filtering functions, dsa_port_is_{unused,user,cpu_dsa}, can be used
directly on the iterator. This eliminates the second lookup.

These functions can be used both by the core and by drivers.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: default avatarFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      82b31898
    • David S. Miller's avatar
      Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue · dedb0809
      David S. Miller authored 
      Tony Nguyen says:

====================
100GbE Intel Wired LAN Driver Updates 2021-10-20

Sudheer Mogilappagari says:

This series introduces initial support for Application Device Queues(ADQ)
in ice driver. ADQ provides traffic isolation for application flows in
hardware and ability to steer traffic to a given traffic class. This
helps in aligning NIC queues to application threads.

Traffic classes are configured using mqprio framework of tc command
and mapped to HW channels(VSIs) in the driver. The queue set of each
traffic class is managed by corresponding VSI. Each traffic channel
can be configured with bandwidth rate-limiting limits and is offloaded
to the hardware through the mqprio framework by specifying the mode
option as 'channel' and shaper option as 'bw_rlimit'.

Next, the flows of application can be steered into a given traffic class
using "tc filter" command. The option "skip_sw hw_tc x" indicates
hw-offload of filtering and steering filtered traffic into specified TC.
Non-matching traffic flows through TC0.

When channel configuration are removed queue configuration is set to
default and filters configured on individual traffic classes are deleted.

example:
$ ethtool -K eth0 hw-tc-offload on

Configure 3 traffic classes and map priority 0,1,2 to TC0, TC1 and TC2
respectively. TC0 has 2 queues from offset 0 & TC1 has 8 queues from
offset 2 and TC2 has 4 queues from offset 10. Enable hardware offload
of channels.

$ tc qdisc add dev eth0 root mqprio num_tc 3 map 0 1 2 queues \
        2@0 8@2 4@10 hw 1 mode channel

$ tc qdisc show dev eth0
qdisc mqprio 8001: root  tc 2 map 0 1 2 0 0 0 0 0 0 0 0 0 0 0 0 0
             queues:(0:1) (2:9) (10:13)
             mode:channel

Configure two filters to match based on dst ipaddr, dst tcp port and
redirect to TC1 and TC2.
$ tc qdisc add dev eth0 clsact

$ tc filter add dev eth0 protocol ip ingress prio 1 flower\
  dst_ip 192.168.1.1/32 ip_proto tcp dst_port 80\
  skip_sw hw_tc 1
$ tc filter add dev eth0 protocol ip ingress prio 1 flower\
  dst_ip 192.168.1.1/32 ip_proto tcp dst_port 5001\
  skip_sw hw_tc 2

$ tc filter show dev eth0 ingress

Delete traffic classes configuration:
$ sudo tc qdisc del dev eth0 root
====================
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      dedb0809
    • David S. Miller's avatar
      Merge branch 'mscc-ocelot-all-ports-vlan-untagged-egress' · 7d4f4d14
      David S. Miller authored 
      Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      7d4f4d14
    • Vladimir Oltean's avatar
      net: mscc: ocelot: track the port pvid using a pointer · d4004422
      Vladimir Oltean authored 
      Now that we have a list of struct ocelot_bridge_vlan entries, we can
rewrite the pvid logic to simply point to one of those structures,
instead of having a separate structure with a "bool valid".
The NULL pointer will represent the lack of a bridge pvid (not to be
confused with the lack of a hardware pvid on the port, that is present
at all times).
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      d4004422
    • Vladimir Oltean's avatar
      net: mscc: ocelot: add the local station MAC addresses in VID 0 · bfbab310
      Vladimir Oltean authored 
      The ocelot switchdev driver does not include the CPU port in the list of
flooding destinations for unknown traffic, instead that traffic is
supposed to match FDB entries to reach the CPU.

The addresses it installs are:
(a) the station MAC address, in ocelot_probe_port() and later during
    runtime in ocelot_port_set_mac_address(). These are the VLAN-unaware
    addresses. The VLAN-aware addresses are in ocelot_vlan_vid_add().
(b) multicast addresses added with dev_mc_add() (not bridge host MDB
    entries) in ocelot_mc_sync()
(c) multicast destination MAC addresses for MRP in ocelot_mrp_save_mac(),
    to make sure those are dropped (not forwarded) by the bridging
    service, just trapped to the CPU

So we can see that the logic is slightly buggy ever since the initial
commit a556c76a ("net: mscc: Add initial Ocelot switch support").
This is because, when ocelot_probe_port() runs, the port pvid is 0.
Then we join a VLAN-aware bridge, the pvid becomes 1, we call
ocelot_port_set_mac_address(), this learns the new MAC address in VID 1
(also fails to forget the old one, since it thinks it's in VID 1, but
that's not so important). Then when we leave the VLAN-aware bridge,
outside world is unable to ping our new MAC address because it isn't
learned in VID 0, the VLAN-unaware pvid.

[ note: this is strictly based on static analysis, I don't have hardware
  to test. But there are also many more corner cases ]

The basic idea is that we should have a separation of concerns, and the
FDB entries used for standalone operation should be managed by the
driver, and the FDB entries used by the bridging service should be
managed by the bridge. So the standalone and VLAN-unaware bridge FDB
entries should not follow the bridge PVID, because that will only be
active when the bridge is VLAN-aware. So since the port pvid is
coincidentally zero during probe time, just make those entries
statically go to VID 0.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      bfbab310
    • Vladimir Oltean's avatar
      net: mscc: ocelot: allow a config where all bridge VLANs are egress-untagged · 0da1a1c4
      Vladimir Oltean authored 
      At present, the ocelot driver accepts a single egress-untagged bridge
VLAN, meaning that this sequence of operations:

ip link add br0 type bridge vlan_filtering 1
ip link set swp0 master br0
bridge vlan add dev swp0 vid 2 pvid untagged

fails because the bridge automatically installs VID 1 as a pvid & untagged
VLAN, and vid 2 would be the second untagged VLAN on this port. It is
necessary to delete VID 1 before proceeding to add VID 2.

This limitation comes from the fact that we operate the port tag, when
it has an egress-untagged VID, in the OCELOT_PORT_TAG_NATIVE mode.
The ocelot switches do not have full flexibility and can either have one
single VID as egress-untagged, or all of them.

There are use cases for having all VLANs as egress-untagged as well, and
this patch adds support for that.

The change rewrites ocelot_port_set_native_vlan() into a more generic
ocelot_port_manage_port_tag() function. Because the software bridge's
state, transmitted to us via switchdev, can become very complex, we
don't attempt to track all possible state transitions, but instead take
a more declarative approach and just make ocelot_port_manage_port_tag()
figure out which more to operate in:

- port is VLAN-unaware: the classified VLAN (internal, unrelated to the
                        802.1Q header) is not inserted into packets on egress
- port is VLAN-aware:
  - port has tagged VLANs:
    -> port has no untagged VLAN: set up as pure trunk
    -> port has one untagged VLAN: set up as trunk port + native VLAN
    -> port has more than one untagged VLAN: this is an invalid config
       which is rejected by ocelot_vlan_prepare
  - port has no tagged VLANs
    -> set up as pure egress-untagged port

We don't keep the number of tagged and untagged VLANs, we just count the
structures we keep.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      0da1a1c4
    • Vladimir Oltean's avatar
      net: mscc: ocelot: convert the VLAN masks to a list · 90e0aa8d
      Vladimir Oltean authored 
      First and foremost, the driver currently allocates a constant sized
4K * u32 (16KB memory) array for the VLAN masks. However, a typical
application might not need so many VLANs, so if we dynamically allocate
the memory as needed, we might actually save some space.

Secondly, we'll need to keep more advanced bookkeeping of the VLANs we
have, notably we'll have to check how many untagged and how many tagged
VLANs we have. This will have to stay in a structure, and allocating
another 16 KB array for that is again a bit too much.

So refactor the bridge VLANs in a linked list of structures.

The hook points inside the driver are ocelot_vlan_member_add() and
ocelot_vlan_member_del(), which previously used to operate on the
ocelot->vlan_mask[vid] array element.

ocelot_vlan_member_add() and ocelot_vlan_member_del() used to call
ocelot_vlan_member_set() to commit to the ocelot->vlan_mask.
Additionally, we had two calls to ocelot_vlan_member_set() from outside
those callers, and those were directly from ocelot_vlan_init().
Those calls do not set up bridging service VLANs, instead they:

- clear the VLAN table on reset
- set the port pvid to the value used by this driver for VLAN-unaware
  standalone port operation (VID 0)

So now, when we have a structure which represents actual bridge VLANs,
VID 0 doesn't belong in that structure, since it is not part of the
bridging layer.

So delete the middle man, ocelot_vlan_member_set(), and let
ocelot_vlan_init() call directly ocelot_vlant_set_mask() which forgoes
any data structure and writes directly to hardware, which is all that we
need.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      90e0aa8d
    • Vladimir Oltean's avatar
      net: mscc: ocelot: add a type definition for REW_TAG_CFG_TAG_CFG · 62a22bcb
      Vladimir Oltean authored 
      This is a cosmetic patch which clarifies what are the port tagging
options for Ocelot switches.
Signed-off-by: default avatarVladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      62a22bcb
  3. 20 Oct, 2021 25 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7