David Wei says:

====================
netdevsim: link and forward skbs between ports

This patchset adds the ability to link two netdevsim ports together and
forward skbs between them, similar to veth. The goal is to use netdevsim
for testing features e.g. zero copy Rx using io_uring.

This feature was tested locally on QEMU, and a selftest is included.

I ran netdev selftests CI style and all tests but the following passed:
- gro.sh
- l2tp.sh
- ip_local_port_range.sh

gro.sh fails because virtme-ng mounts as read-only and it tries to write
to log.txt. This issue was reported to virtme-ng upstream.

l2tp.sh and ip_local_port_range.sh both fail for me on net-next/main as
well.

---
v13->v14:
- implement ndo_get_iflink()
- fix returning 0 if peer is already linked during linking or not linked
  during unlinking
- bump dropped counter if nsim_ipsec_tx() fails and generally reorder
  nsim_start_xmit()
- fix overflowing lines and indentations

v12->v13:
- wait for socat listening port to be ready before sending data in
  selftest

v11->v12:
- fix leaked netns refs
- fix rtnetlink.sh kci_test_ipsec_offload() selftest

v10->v11:
- add udevadm settle after creating netdevsims in selftest

v9->v10:
- fix not freeing skb when not there is no peer
- prevent possible id clashes in selftest
- cleanup selftest on error paths

v8->v9:
- switch to getting netns using fd rather than id
- prevent linking a netdevsim to itself
- update tests

v7->v8:
- fix not dereferencing RCU ptr using rcu_dereference()
- remove unused variables in selftest

v6->v7:
- change link syntax to netnsid:ifidx
- replace dev_get_by_index() with __dev_get_by_index()
- check for NULL peer when linking
- add a sysfs attribute for unlinking
- only update Tx stats if not dropped
- update selftest

v5->v6:
- reworked to link two netdevsims using sysfs attribute on the bus
  device instead of debugfs due to deadlock possibility if a netdevsim
  is removed during linking
- removed unnecessary patch maintaining a list of probed nsim_devs
- updated selftest

v4->v5:
- reduce nsim_dev_list_lock critical section
- fixed missing mutex unlock during unwind ladder
- rework nsim_dev_peer_write synchronization to take devlink lock as
  well as rtnl_lock
- return err msgs to user during linking if port doesn't exist or
  linking to self
- update tx stats outside of RCU lock

v3->v4:
- maintain a mutex protected list of probed nsim_devs instead of using
  nsim_bus_dev
- fixed synchronization issues by taking rtnl_lock
- track tx_dropped skbs

v2->v3:
- take lock when traversing nsim_bus_dev_list
- take device ref when getting a nsim_bus_dev
- return 0 if nsim_dev_peer_read cannot find the port
- address code formatting
- do not hard code values in selftests
- add Makefile for selftests

v1->v2:
- renamed debugfs file from "link" to "peer"
- replaced strstep() with sscanf() for consistency
- increased char[] buf sz to 22 for copying id + port from user
- added err msg w/ expected fmt when linking as a hint to user
- prevent linking port to itself
- protect peer ptr using RCU

====================
Signed-off-by: David S. Miller <davem@davemloft.net>

I cleared IFF_NOARP flag from netdevsim dev->flags in order to support
skb forwarding. This breaks the rtnetlink.sh selftest
kci_test_ipsec_offload() test because ipsec does not connect to peers it
cannot transmit to.

Fix the issue by adding a neigh entry manually. ipsec_offload test now
successfully pass.
Signed-off-by: David Wei <dw@davidwei.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

Connect two netdevsim ports in different namespaces together, then send
packets between them using socat.
Signed-off-by: David Wei <dw@davidwei.uk>
Reviewed-by: Maciek Machnikowski <maciek@machnikowski.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add an implementation for ndo_get_iflink() in netdevsim that shows the
ifindex of the linked peer, if any.
Signed-off-by: David Wei <dw@davidwei.uk>
Reviewed-by: Maciek Machnikowski <maciek@machnikowski.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Forward skbs sent from one netdevsim port to its connected netdevsim
port using dev_forward_skb, in a spirit similar to veth.

Add a tx_dropped variable to struct netdevsim, tracking the number of
skbs that could not be forwarded using dev_forward_skb().

The xmit() function accessing the peer ptr is protected by an RCU read
critical section. The rcu_read_lock() is functionally redundant as since
v5.0 all softirqs are implicitly RCU read critical sections; but it is
useful for human readers.

If another CPU is concurrently in nsim_destroy(), then it will first set
the peer ptr to NULL. This does not affect any existing readers that
dereferenced a non-NULL peer. Then, in unregister_netdevice(), there is
a synchronize_rcu() before the netdev is actually unregistered and
freed. This ensures that any readers i.e. xmit() that got a non-NULL
peer will complete before the netdev is freed.

Any readers after the RCU_INIT_POINTER() but before synchronize_rcu()
will dereference NULL, making it safe.

The codepath to nsim_destroy() and nsim_create() takes both the newly
added nsim_dev_list_lock and rtnl_lock. This makes it safe with
concurrent calls to linking two netdevsims together.
Signed-off-by: David Wei <dw@davidwei.uk>
Reviewed-by: Maciek Machnikowski <maciek@machnikowski.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add two netdevsim bus attribute to sysfs:
/sys/bus/netdevsim/link_device
/sys/bus/netdevsim/unlink_device

Writing "A M B N" to link_device will link netdevsim M in netnsid A with
netdevsim N in netnsid B.

Writing "A M" to unlink_device will unlink netdevsim M in netnsid A from
its peer, if any.

rtnl_lock is taken to ensure nothing changes during the linking.
Signed-off-by: David Wei <dw@davidwei.uk>
Reviewed-by: Maciek Machnikowski <maciek@machnikowski.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jakub Kicinski says:

====================
selftests: kselftest_harness: support using xfail

When running selftests for our subsystem in our CI we'd like all
tests to pass. Currently some tests use SKIP for cases they
expect to fail, because the kselftest_harness limits the return
codes to pass/fail/skip. XFAIL which would be a great match
here cannot be used.

Remove the no_print handling and use vfork() to run the test in
a different process than the setup. This way we don't need to
pass "failing step" via the exit code. Further clean up the exit
codes so that we can use all KSFT_* values. Rewrite the result
printing to make handling XFAIL/XPASS easier. Support tests
declaring combinations of fixture + variant they expect to fail.

Merge plan is to put it on top of -rc6 and merge into net-next.
That way others should be able to pull the patches without
any networking changes.

v4:
 - rebase on top of Mickael's vfork() changes
v3: https://lore.kernel.org/all/20240220192235.2953484-1-kuba@kernel.org/
 - combine multiple series
 - change to "list of expected failures" rather than SKIP()-like handling
v2: https://lore.kernel.org/all/20240216002619.1999225-1-kuba@kernel.org/
 - fix alignment
follow up RFC: https://lore.kernel.org/all/20240216004122.2004689-1-kuba@kernel.org/
v1: https://lore.kernel.org/all/20240213154416.422739-1-kuba@kernel.org/
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

SCTP does not support IP_LOCAL_PORT_RANGE and we know it,
so use XFAIL instead of SKIP.
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently some tests report skip for things they expect to fail
e.g. when given combination of parameters is known to be unsupported.
This is confusing because in an ideal test environment and fully
featured kernel no tests should be skipped.

Selftest summary line already includes xfail and xpass counters,
e.g.:

  Totals: pass:725 fail:0 xfail:0 xpass:0 skip:0 error:0

but there's no way to use it from within the harness.

Add a new per-fixture+variant combination list of test cases
we expect to fail.
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Switch to printing KTAP line for PASS / FAIL with ksft_test_result_code(),
this gives us the ability to report diagnostic messages.
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

According to the spec we should always print a # if we add
a diagnostic message. Having the caller pass in the new line
as part of diagnostic message makes handling this a bit
counter-intuitive, so append the new line in the helper.
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jakub points out that for parsers it's rather useful to always
have the test name on the result line. Currently if we SKIP
(or soon XFAIL or XPASS), we will print:

ok 17 # SKIP SCTP doesn't support IP_BIND_ADDRESS_NO_PORT

     ^
     no test name

Always print the test name.
KTAP format seems to allow or even call for it, per:
https://docs.kernel.org/dev-tools/ktap.htmlSuggested-by: Jakub Sitnicki <jakub@cloudflare.com>
Link: https://lore.kernel.org/all/87jzn6lnou.fsf@cloudflare.com/Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

For generic test harness code it's more useful to deal with exit
codes directly, rather than having to switch on them and call
the right ksft_test_result_*() helper. Add such function to kselftest.h.

Note that "directive" and "diagnostic" are what ktap docs call
those parts of the message.
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

We always use skip in combination with exit_code being 0
(KSFT_PASS). This are basic KSFT / KTAP semantics.
Store the right KSFT_* code in exit_code directly.

This makes it easier to support tests reporting other
extended KSFT_* codes like XFAIL / XPASS.
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Instead of tracking passed = 0/1 rename the field to exit_code
and invert the values so that they match the KSFT_* exit codes.
This will allow us to fold SKIP / XFAIL into the same value.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Since we added variant support generating full test case
name takes 4 string arguments. We're about to need it
in another two places. Stop the duplication and print
once into a temporary buffer.
Suggested-by: Jakub Sitnicki <jakub@cloudflare.com>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Now that we no longer need low exit codes to communicate
assertion steps - use normal KSFT exit codes.
Acked-by: Kees Cook <keescook@chromium.org>
Tested-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Replace Landlock-specific TEST_F_FORK() with an improved TEST_F() which
brings four related changes:

Run TEST_F()'s tests in a grandchild process to make it possible to
drop privileges and delegate teardown to the parent.

Compared to TEST_F_FORK(), simplify handling of the test grandchild
process thanks to vfork(2), and makes it generic (e.g. no explicit
conversion between exit code and _metadata).

Compared to TEST_F_FORK(), run teardown even when tests failed with an
assert thanks to commit 63e6b2a4 ("selftests/harness: Run TEARDOWN
for ASSERT failures").

Simplify the test harness code by removing the no_print and step fields
which are not used.  I added this feature just after I made
kselftest_harness.h more broadly available but this step counter
remained even though it wasn't needed after all. See commit 369130b6
("selftests: Enhance kselftest_harness.h to print which assert failed").

Replace spaces with tabs in one line of __TEST_F_IMPL().

Cc: Günther Noack <gnoack@google.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Will Drewry <wad@chromium.org>
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This has the effect of creating a new test process for either TEST_F()
or TEST_F_FORK(), which doesn't change tests but will ease potential
backports.  See next commit for the TEST_F_FORK() merge into TEST_F().

Cc: Günther Noack <gnoack@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Will Drewry <wad@chromium.org>
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Justin Chen says:

====================
Support for ASP 2.2 and optimizations

ASP 2.2 adds some power savings during low power modes.

Also make various improvements when entering low power modes and
reduce MDIO traffic by hooking up interrupts.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Hook up the phy interrupts for internal phys to reduce mdio traffic
and improve responsiveness of link changes.
Signed-off-by: Justin Chen <justin.chen@broadcom.com>
Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There is no advantage of freeing and re-allocating buffers through
suspend and resume. This waste cycles and makes suspend/resume time
longer. We also open ourselves to failed allocations in systems with
heavy memory fragmentation.
Signed-off-by: Justin Chen <justin.chen@broadcom.com>
Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add mdio compat string for ASP 2.0 ethernet driver.
Signed-off-by: Justin Chen <justin.chen@broadcom.com>
Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ASP 2.2 improves power savings during low power modes.

A new register was added to toggle to a slower clock during low
power modes.

EEE was broken for ASP 2.0/2.1. A HW workaround was added for
ASP 2.2 that requires toggling a chicken bit.
Signed-off-by: Justin Chen <justin.chen@broadcom.com>
Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add support for ASP 2.2.
Signed-off-by: Justin Chen <justin.chen@broadcom.com>
Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The ASP 2.2 Ethernet controller uses a brcm unimac.
Signed-off-by: Justin Chen <justin.chen@broadcom.com>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Robert Marko says:

====================
net: phy: qcom: qca808x: fill in possible_interfaces

QCA808x does not currently fill in the possible_interfaces.

This leads to Phylink not being aware that it supports 2500Base-X as well
so in cases where it is connected to a DSA switch like MV88E6393 it will
limit that port to phy-mode set in the DTS.

That means that if SGMII is used you are limited to 1G only while if
2500Base-X was set you are limited to 2.5G only.

Populating the possible_interfaces fixes this.

Changes in v2:
* Get rid of the if/else by Russels suggestion in the helper
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently QCA808x driver does not fill the possible_interfaces.
2.5G QCA808x support SGMII and 2500Base-X while 1G model only supports
SGMII, so fill the possible_interfaces accordingly.
Signed-off-by: Robert Marko <robimarko@gmail.com>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

There are 2 versions of QCA808x, one 2.5G capable and one 1G capable.
Currently, this matter only in the .get_features call however, it will
be required for filling supported interface modes so lets add a helper
that can be reused.
Signed-off-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There is no need to wrap calls to the no_printk() helper inside an
always-false check, as no_printk() already does that internally.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Eric Dumazet says:

====================
ipv6: lockless accesses to devconf

- First patch puts in a cacheline_group the fields used in fast paths.

- Annotate all data races around idev->cnf fields.

- Last patch in this series removes RTNL use for RTM_GETNETCONF dumps.

v3: addressed Jakub Kicinski feedback in addrconf_disable_ipv6()
    Added tags from Jiri and Florian.

v2: addressed Jiri Pirko feedback
 - Added "ipv6: addrconf_disable_ipv6() optimizations"
   and "ipv6: addrconf_disable_policy() optimization"
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

1) inet6_netconf_dump_devconf() can run under RCU protection
   instead of RTNL.

2) properly return 0 at the end of a dump, avoiding an
   an extra recvmsg() system call.

3) Do not use inet6_base_seq() anymore, for_each_netdev_dump()
   has nice properties. Restarting a GETDEVCONF dump if a device has
   been added/removed or if net->ipv6.dev_addr_genid has changed is moot.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Final (?) round of this series.

Annotate lockless reads on following devconf fields,
because they be changed concurrently from /proc/net/ipv6/conf.

- accept_dad
- optimistic_dad
- use_optimistic
- use_oif_addrs_only
- ra_honor_pio_life
- keep_addr_on_down
- ndisc_notify
- ndisc_evict_nocarrier
- suppress_frag_ndisc
- addr_gen_mode
- seg6_enabled
- ioam6_enabled
- ioam6_id
- ioam6_id_wide
- drop_unicast_in_l2_multicast
- mldv[12]_unsolicited_report_interval
- force_mld_version
- force_tllao
- accept_untracked_na
- drop_unsolicited_na
- accept_source_route
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Annotate lockless reads and writes on following devconf fields:

- regen_min_advance
- regen_max_retry
- dad_transmits
- use_tempaddr
- max_addresses
- max_desync_factor
- temp_valid_lft
- rtr_solicits
- rtr_solicit_max_interval
- rtr_solicit_interval
- rtr_solicit_delay
- enhanced_dad
- accept_redirects
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Writing over /proc/sys/net/ipv6/conf/default/disable_policy
does not need to hold RTNL.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

idev->cnf.disable_policy and net->ipv6.devconf_all->disable_policy
can be read locklessly. Add appropriate annotations on reads
and writes.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

devconf->proxy_ndp can be read and written locklessly,
add appropriate annotations.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use READ_ONCE() while reading idev->cnf.rtr_probe_interval
while its value could be changed.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

idev->cnf.ignore_routes_with_linkdown can be used without any locks,
add appropriate annotations.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Annotate reads from in6_dev->cnf.XXX fields, as they could
change concurrently.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>