platform/kernel/linux-starfive.git
4 years agonet: dsa: sja1105: add configuration of port policers
Vladimir Oltean [Sun, 29 Mar 2020 11:52:01 +0000 (14:52 +0300)]
net: dsa: sja1105: add configuration of port policers

This adds partial configuration support for the L2 Policing Table. Out
of the 45 policing entries, only 5 are used (one for each port), in a
shared manner. All 8 traffic classes, and the broadcast policer, are
redirected to a common instance which belongs to the ingress port.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: dsa: felix: add port policers
Vladimir Oltean [Sun, 29 Mar 2020 11:52:00 +0000 (14:52 +0300)]
net: dsa: felix: add port policers

This patch is a trivial passthrough towards the ocelot library, which
support port policers since commit 2c1d029a017f ("net: mscc: ocelot:
Implement port policers via tc command").

Some data structure conversion between the DSA core and the Ocelot
library is necessary, for policer parameters.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: dsa: add port policers
Vladimir Oltean [Sun, 29 Mar 2020 11:51:59 +0000 (14:51 +0300)]
net: dsa: add port policers

The approach taken to pass the port policer methods on to drivers is
pragmatic. It is similar to the port mirroring implementation (in that
the DSA core does all of the filter block interaction and only passes
simple operations for the driver to implement) and dissimilar to how
flow-based policers are going to be implemented (where the driver has
full control over the flow_cls_offload data structure).

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: dsa: refactor matchall mirred action to separate function
Vladimir Oltean [Sun, 29 Mar 2020 11:51:58 +0000 (14:51 +0300)]
net: dsa: refactor matchall mirred action to separate function

Make room for other actions for the matchall filter by keeping the
mirred argument parsing self-contained in its own function.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: mscc: ocelot: add action of police on vcap_is2
Xiaoliang Yang [Sun, 29 Mar 2020 11:51:57 +0000 (14:51 +0300)]
net: mscc: ocelot: add action of police on vcap_is2

Ocelot has 384 policers that can be allocated to ingress ports,
QoS classes per port, and VCAP IS2 entries. ocelot_police.c
supports to set policers which can be allocated to police action
of VCAP IS2. We allocate policers from maximum pol_id, and
decrease the pol_id when add a new vcap_is2 entry which is
police action.

Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'ionic-support-for-firmware-upgrade'
David S. Miller [Mon, 30 Mar 2020 18:40:50 +0000 (11:40 -0700)]
Merge branch 'ionic-support-for-firmware-upgrade'

Shannon Nelson says:

====================
ionic support for firmware upgrade

The Pensando Distributed Services Card can get firmware upgrades from
the off-host centralized management suite, and can be upgraded without a
host reboot or driver reload.  This patchset sets up the support for fw
upgrade in the Linux driver.

When the upgrade begins, the DSC first brings the link down, then stops
the firmware.  The driver will notice this and quiesce itself by stopping
the queues and releasing DMA resources, then monitoring for firmware to
start back up.  When the upgrade is finished the firmware is restarted
and link is brought up, and the driver rebuilds the queues and restarts
traffic flow.

First we separate the Link state from the netdev state, then reorganize a
few things to prepare for partial tear-down of the queues.  Next we fix
up the state machine so that we take the Tx and Rx queues down and back
up when we get LINK_DOWN and LINK_UP events.  Lastly, we add handling of
the FW reset itself by tearing down the lif internals and rebuilding them
with the new FW setup.

v2: This changes the design from (ab)using the full .ndo_stop and
    .ndo_open routines to getting a better separation between the
    alloc and the init functions so that we can keep our resource
    allocations as long as possible.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: remove lifs on fw reset
Shannon Nelson [Sat, 28 Mar 2020 03:14:48 +0000 (20:14 -0700)]
ionic: remove lifs on fw reset

When the FW RESET event comes to the driver from the firmware,
or the fw_status goes to 0 (stopped) or to 0xff (no PCI
connection), then shut down the driver activity.  This event
signals a FW upgrade where we need to quiesce all operations and
wait for the FW to restart.  The FW will continue the update
process once it sees all the LIFs are reset.  When the update
process is done it will set the fw_status back to RUNNING.
Meanwhile, the heartbeat check continues and when the fw_status
is seen as set to running we can restart the driver operations.

Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: disable the queues on link down
Shannon Nelson [Sat, 28 Mar 2020 03:14:47 +0000 (20:14 -0700)]
ionic: disable the queues on link down

When the link goes down, we need to disable the queues on the
NIC in addition to stopping the netdev stack.  This lets the
FW know that the driver has stopped queue activity, and then
the FW can do internal reconfiguration work, whether actually
Link related, or for other internal FW needs.  To do this,
we pull out the queue enable and disable from ionic_open()
and ionic_stop() so they can be used by other routines.

To help keep things sane, we swap the queue enables so that
the rx queue and its napi are enabled before the tx queue
which rides on the rx queues napi.

We also drop the ionic_lif_quiesce() as it doesn't do anything
more than what the queue disable has already taken care of.

Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: check for queues before deleting
Shannon Nelson [Sat, 28 Mar 2020 03:14:46 +0000 (20:14 -0700)]
ionic: check for queues before deleting

Make sure the queue structures exist before trying
to delete them.  This addresses a couple of error
recovery issues.

Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: clean tx queue of unfinished requests
Shannon Nelson [Sat, 28 Mar 2020 03:14:45 +0000 (20:14 -0700)]
ionic: clean tx queue of unfinished requests

Clean out tx requests that didn't get finished before
shutting down the queue.

Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: move irq request to qcq alloc
Shannon Nelson [Sat, 28 Mar 2020 03:14:44 +0000 (20:14 -0700)]
ionic: move irq request to qcq alloc

Move the irq request and free out of the qcq_init and deinit
and into the alloc and free routines where they belong for
better resource management.

Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: move debugfs add/delete to match alloc/free
Shannon Nelson [Sat, 28 Mar 2020 03:14:43 +0000 (20:14 -0700)]
ionic: move debugfs add/delete to match alloc/free

Move the qcq debugfs add to the queue alloc, and likewise move
the debugfs delete to the queue free.  The LIF debugfs add
also needs to be moved, but the del is already in the LIF free.

Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: check for linkup in watchdog
Shannon Nelson [Sat, 28 Mar 2020 03:14:42 +0000 (20:14 -0700)]
ionic: check for linkup in watchdog

Add a link_status_check to the heartbeat watchdog.

Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: decouple link message from netdev state
Shannon Nelson [Sat, 28 Mar 2020 03:14:41 +0000 (20:14 -0700)]
ionic: decouple link message from netdev state

Rearrange the link_up/link_down messages so that we announce
link up when we first notice that the link is up when the
driver loads, and decouple the link_up/link_down messages from
the UP and DOWN netdev state.

Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_ptp: Fix build warnings
Ido Schimmel [Mon, 30 Mar 2020 18:08:20 +0000 (21:08 +0300)]
mlxsw: spectrum_ptp: Fix build warnings

Cited commit extended the enums 'hwtstamp_tx_types' and
'hwtstamp_rx_filters' with values that were not accounted for in the
switch statements, resulting in the build warnings below.

Fix by adding a default case.

drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.c: In function ‘mlxsw_sp_ptp_get_message_types’:
drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.c:915:2: warning: enumeration value ‘__HWTSTAMP_TX_CNT’ not handled in switch [-Wswitch]
  915 |  switch (tx_type) {
      |  ^~~~~~
drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.c:927:2: warning: enumeration value ‘__HWTSTAMP_FILTER_CNT’ not handled in switch [-Wswitch]
  927 |  switch (rx_filter) {
      |  ^~~~~~

Fixes: f76510b458a5 ("ethtool: add timestamping related string sets")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reported-by: David S. Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'Devlink-health-auto-attributes-refactor'
David S. Miller [Mon, 30 Mar 2020 18:17:40 +0000 (11:17 -0700)]
Merge branch 'Devlink-health-auto-attributes-refactor'

Eran Ben Elisha says:

====================
Devlink health auto attributes refactor

This patchset refactors the auto-recover health reporter flag to be
explicitly set by the devlink core.
In addition, add another flag to control auto-dump attribute, also
to be explicitly set by the devlink core.

For that, patch 0001 changes the auto-recover default value of
netdevsim dummy reporter.

After reporter registration, both flags can be altered be administrator
only.

Changes since v1:
- Change default behaviour of netdevsim dummy reporter
- Move initialization of DEVLINK_ATTR_HEALTH_REPORTER_AUTO_DUMP
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agodevlink: Add auto dump flag to health reporter
Eran Ben Elisha [Sun, 29 Mar 2020 11:05:55 +0000 (14:05 +0300)]
devlink: Add auto dump flag to health reporter

On low memory system, run time dumps can consume too much memory. Add
administrator ability to disable auto dumps per reporter as part of the
error flow handle routine.

This attribute is not relevant while executing
DEVLINK_CMD_HEALTH_REPORTER_DUMP_GET.

By default, auto dump is activated for any reporter that has a dump method,
as part of the reporter registration to devlink.

Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agodevlink: Implicitly set auto recover flag when registering health reporter
Eran Ben Elisha [Sun, 29 Mar 2020 11:05:54 +0000 (14:05 +0300)]
devlink: Implicitly set auto recover flag when registering health reporter

When health reporter is registered to devlink, devlink will implicitly set
auto recover if and only if the reporter has a recover method. No reason
to explicitly get the auto recover flag from the driver.

Remove this flag from all drivers that called
devlink_health_reporter_create.

All existing health reporters set auto recovery to true if they have a
recover method.

Yet, administrator can unset auto recover via netlink command as prior to
this patch.

Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonetdevsim: Change dummy reporter auto recover default
Eran Ben Elisha [Sun, 29 Mar 2020 11:05:53 +0000 (14:05 +0300)]
netdevsim: Change dummy reporter auto recover default

Health reporters should be registered with auto recover set to true.
Align dummy reporter behaviour with that, as in later patch the option to
set auto recover behaviour will be removed.

In addition, align netdevsim selftest to the new default value.

Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoptp: Avoid deadlocks in the programmable pin code.
Richard Cochran [Sun, 29 Mar 2020 14:55:10 +0000 (07:55 -0700)]
ptp: Avoid deadlocks in the programmable pin code.

The PTP Hardware Clock (PHC) subsystem offers an API for configuring
programmable pins.  User space sets or gets the settings using ioctls,
and drivers verify dialed settings via a callback.  Drivers may also
query pin settings by calling the ptp_find_pin() method.

Although the core subsystem protects concurrent access to the pin
settings, the implementation places illogical restrictions on how
drivers may call ptp_find_pin().  When enabling an auxiliary function
via the .enable(on=1) callback, drivers may invoke the pin finding
method, but when disabling with .enable(on=0) drivers are not
permitted to do so.  With the exception of the mv88e6xxx, all of the
PHC drivers do respect this restriction, but still the locking pattern
is both confusing and unnecessary.

This patch changes the locking implementation to allow PHC drivers to
freely call ptp_find_pin() from their .enable() and .verify()
callbacks.

V2 ChangeLog:
- fixed spelling in the kernel doc
- add Vladimir's tested by tag

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Reported-by: Yangbo Lu <yangbo.lu@nxp.com>
Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: devlink: use NL_SET_ERR_MSG_MOD instead of NL_SET_ERR_MSG
Jiri Pirko [Sat, 28 Mar 2020 18:25:29 +0000 (19:25 +0100)]
net: devlink: use NL_SET_ERR_MSG_MOD instead of NL_SET_ERR_MSG

The rest of the devlink code sets the extack message using
NL_SET_ERR_MSG_MOD. Change the existing appearances of NL_SET_ERR_MSG
to NL_SET_ERR_MSG_MOD.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'net-sched-expose-HW-stats-types-per-action-used-by-drivers'
David S. Miller [Mon, 30 Mar 2020 18:06:49 +0000 (11:06 -0700)]
Merge branch 'net-sched-expose-HW-stats-types-per-action-used-by-drivers'

Jiri Pirko says:

====================
net: sched: expose HW stats types per action used by drivers

The first patch is just adding a helper used by the second patch too.
The second patch is exposing HW stats types that are used by drivers.

Example:

$ tc filter add dev enp3s0np1 ingress proto ip handle 1 pref 1 flower dst_ip 192.168.1.1 action drop
$ tc -s filter show dev enp3s0np1 ingress
filter protocol ip pref 1 flower chain 0
filter protocol ip pref 1 flower chain 0 handle 0x1
  eth_type ipv4
  dst_ip 192.168.1.1
  in_hw in_hw_count 2
        action order 1: gact action drop
         random type none pass val 0
         index 1 ref 1 bind 1 installed 10 sec used 10 sec
        Action statistics:
        Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
        backlog 0b 0p requeues 0
        used_hw_stats immediate     <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: sched: expose HW stats types per action used by drivers
Jiri Pirko [Sat, 28 Mar 2020 15:37:43 +0000 (16:37 +0100)]
net: sched: expose HW stats types per action used by drivers

It may be up to the driver (in case ANY HW stats is passed) to select
which type of HW stats he is going to use. Add an infrastructure to
expose this information to user.

$ tc filter add dev enp3s0np1 ingress proto ip handle 1 pref 1 flower dst_ip 192.168.1.1 action drop
$ tc -s filter show dev enp3s0np1 ingress
filter protocol ip pref 1 flower chain 0
filter protocol ip pref 1 flower chain 0 handle 0x1
  eth_type ipv4
  dst_ip 192.168.1.1
  in_hw in_hw_count 2
        action order 1: gact action drop
         random type none pass val 0
         index 1 ref 1 bind 1 installed 10 sec used 10 sec
        Action statistics:
        Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
        backlog 0b 0p requeues 0
        used_hw_stats immediate     <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: introduce nla_put_bitfield32() helper and use it
Jiri Pirko [Sat, 28 Mar 2020 15:37:42 +0000 (16:37 +0100)]
net: introduce nla_put_bitfield32() helper and use it

Introduce a helper to pass value and selector to. The helper packs them
into struct and puts them into netlink message.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec...
David S. Miller [Mon, 30 Mar 2020 17:59:20 +0000 (10:59 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/klassert/ipsec-next

Steffen Klassert says:

====================
pull request (net-next): ipsec-next 2020-03-28

1) Use kmem_cache_zalloc() instead of kmem_cache_alloc()
   in xfrm_state_alloc(). From Huang Zijiang.

2) esp_output_fill_trailer() is the same in IPv4 and IPv6,
   so share this function to avoide code duplcation.
   From Raed Salem.

3) Add offload support for esp beet mode.
   From Xin Long.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: dsa: Simplify 'dsa_tag_protocol_to_str()'
Christophe JAILLET [Sat, 28 Mar 2020 09:53:09 +0000 (10:53 +0100)]
net: dsa: Simplify 'dsa_tag_protocol_to_str()'

There is no point in preparing the module name in a buffer. The format
string can be passed diectly to 'request_module()'.

This axes a few lines of code and cleans a few things:
   - max len for a driver name is MODULE_NAME_LEN wich is ~ 60 chars,
     not 128. It would be down-sized in 'request_module()'
   - we should pass the total size of the buffer to 'snprintf()', not the
     size minus 1

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: ena: Make some functions static
YueHaibing [Sat, 28 Mar 2020 03:06:20 +0000 (11:06 +0800)]
net: ena: Make some functions static

Fix sparse warnings:

drivers/net/ethernet/amazon/ena/ena_netdev.c:460:6: warning: symbol 'ena_xdp_exchange_program_rx_in_range' was not declared. Should it be static?
drivers/net/ethernet/amazon/ena/ena_netdev.c:481:6: warning: symbol 'ena_xdp_exchange_program' was not declared. Should it be static?
drivers/net/ethernet/amazon/ena/ena_netdev.c:1555:5: warning: symbol 'ena_xdp_handle_buff' was not declared. Should it be static?

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agodpaa_eth: Make dpaa_a050385_wa static
YueHaibing [Sat, 28 Mar 2020 03:04:15 +0000 (11:04 +0800)]
dpaa_eth: Make dpaa_a050385_wa static

Fix sparse warning:

drivers/net/ethernet/freescale/dpaa/dpaa_eth.c:2065:5:
 warning: symbol 'dpaa_a050385_wa' was not declared. Should it be static?

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agocrypto/chtls: Fix chtls crash in connection cleanup
Rohit Maheshwari [Mon, 30 Mar 2020 16:55:55 +0000 (22:25 +0530)]
crypto/chtls: Fix chtls crash in connection cleanup

There is a possibility that cdev is removed before CPL_ABORT_REQ_RSS
is fully processed, so it's better to save it in skb.

Added checks in handling the flow correctly, which suggests connection reset
request is sent to HW, wait for HW to respond.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agocrypto/chcr: fix incorrect ipv6 packet length
Rohit Maheshwari [Mon, 30 Mar 2020 16:11:22 +0000 (21:41 +0530)]
crypto/chcr: fix incorrect ipv6 packet length

IPv6 header's payload length field shouldn't include IPv6 header length.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: stmmac: Add support for VLAN Rx filtering
Wong Vee Khee [Mon, 30 Mar 2020 15:53:57 +0000 (23:53 +0800)]
net: stmmac: Add support for VLAN Rx filtering

Add support for VLAN ID-based filtering by the MAC controller for MAC
drivers that support it. Only the 12-bit VID field is used.

Signed-off-by: Chuah Kim Tatt <kim.tatt.chuah@intel.com>
Signed-off-by: Ong Boon Leong <boon.leong.ong@intel.com>
Signed-off-by: Wong Vee Khee <vee.khee.wong@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'crypto-chelsio-Fixes-issues-during-chcr-driver-registration'
David S. Miller [Mon, 30 Mar 2020 17:33:23 +0000 (10:33 -0700)]
Merge branch 'crypto-chelsio-Fixes-issues-during-chcr-driver-registration'

Ayush Sawal says:

====================
crypto: chelsio: Fixes issues during chcr driver registration

Patch 1: Avoid the accessing of wrong u_ctx pointer.
Patch 2: Fixes a deadlock between rtnl_lock and uld_mutex.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoCrypto: chelsio - Fixes a deadlock between rtnl_lock and uld_mutex
Ayush Sawal [Mon, 30 Mar 2020 15:18:53 +0000 (20:48 +0530)]
Crypto: chelsio - Fixes a deadlock between rtnl_lock and uld_mutex

The locks are taken in this order during driver registration
(uld_mutex), at: cxgb4_register_uld.part.14+0x49/0xd60 [cxgb4]
(rtnl_mutex), at: rtnetlink_rcv_msg+0x2db/0x400
(uld_mutex), at: cxgb_up+0x3a/0x7b0 [cxgb4]
(rtnl_mutex), at: chcr_add_xfrmops+0x83/0xa0 [chcr](stucked here)

To avoid this now the netdev features are updated after the
cxgb4_register_uld function is completed.

Fixes: 6dad4e8ab3ec6 ("chcr: Add support for Inline IPSec").

Signed-off-by: Ayush Sawal <ayush.sawal@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoCrypto: chelsio - Fixes a hang issue during driver registration
Ayush Sawal [Mon, 30 Mar 2020 15:18:52 +0000 (20:48 +0530)]
Crypto: chelsio - Fixes a hang issue during driver registration

This issue occurs only when multiadapters are present. Hang
happens because assign_chcr_device returns u_ctx pointer of
adapter which is not yet initialized as for this adapter cxgb_up
is not been called yet.

The last_dev pointer is used to determine u_ctx pointer and it
is initialized two times in chcr_uld_add in chcr_dev_add respectively.

The fix here is don't initialize the last_dev pointer during
chcr_uld_add. Only assign to value to it when the adapter's
initialization is completed i.e in chcr_dev_add.

Fixes: fef4912b66d62 ("crypto: chelsio - Handle PCI shutdown event").

Signed-off-by: Ayush Sawal <ayush.sawal@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoselftests:mptcp: fix failure due to whitespace damage
Matthieu Baerts [Mon, 30 Mar 2020 14:23:54 +0000 (16:23 +0200)]
selftests:mptcp: fix failure due to whitespace damage

'pm_nl_ctl' was adding a trailing whitespace after having printed the
IP. But at the end, the IP element is currently always the last one.

The bash script launching 'pm_nl_ctl' had trailing whitespaces in the
expected result on purpose. But these whitespaces have been removed when
the patch has been applied upstream. To avoid trailing whitespaces in
the bash code, 'pm_nl_ctl' and expected results have now been adapted.

The MPTCP PM selftest can now pass again.

Fixes: eedbc685321b (selftests: add PM netlink functional tests)
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: ethernet: ti: fix spelling mistake "rundom" -> "random"
Colin Ian King [Mon, 30 Mar 2020 10:16:39 +0000 (11:16 +0100)]
net: ethernet: ti: fix spelling mistake "rundom" -> "random"

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

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge tag 'mlx5-updates-2020-03-29' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Mon, 30 Mar 2020 17:14:36 +0000 (10:14 -0700)]
Merge tag 'mlx5-updates-2020-03-29' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5-updates-2020-03-29

1) mlx5 core level updates for mkey APIs + migrate some code to mlx5_ib

2) Use a separate work queue for fib event handling

3) Move new eswitch chains files to a new directory, to prepare for
upcoming E-Switch and offloads features.

4) Support indr block setup (TC_SETUP_FT) in Flow Table mode.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet/mlx5e: add mlx5e_rep_indr_setup_ft_cb support
wenxu [Wed, 25 Mar 2020 12:18:59 +0000 (20:18 +0800)]
net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support

Add mlx5e_rep_indr_setup_ft_cb to support indr block setup
in FT mode.
Both tc rules and flow table rules are of the same format,
It can re-use tc parsing for that, and move the flow table rules
to their steering domain(the specific chain_index), the indr
block offload in FT also follow this scenario.

Signed-off-by: wenxu <wenxu@ucloud.cn>
Reviewed-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5e: refactor indr setup block
wenxu [Wed, 25 Mar 2020 12:18:58 +0000 (20:18 +0800)]
net/mlx5e: refactor indr setup block

Refactor indr setup block for support ft indr setup in the
next patch. The function mlx5e_rep_indr_offload exposes
'flags' in order set additional flag for FT in next patch.
Rename mlx5e_rep_indr_setup_tc_block to mlx5e_rep_indr_setup_block
and add flow_setup_cb_t callback parameters in order set the
specific callback for FT in next patch.

Signed-off-by: wenxu <wenxu@ucloud.cn>
Reviewed-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5: E-Switch: Move eswitch chains to a new directory
Saeed Mahameed [Sat, 14 Mar 2020 01:27:35 +0000 (18:27 -0700)]
net/mlx5: E-Switch: Move eswitch chains to a new directory

eswitch_offloads_chains.{c,h} were just introduced this kernel release
cycle, eswitch is in high development demand right now and many
features are planned to be added to it. eswitch deserves its own
directory and here we move these new files to there, in preparation for
upcoming eswitch features and new files.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
4 years agonet/mlx5: Use a separate work queue for fib event handling
Mark Zhang [Tue, 17 Mar 2020 09:28:28 +0000 (11:28 +0200)]
net/mlx5: Use a separate work queue for fib event handling

In VF lag mode when remove the bonding module without bring down the
bond device first, we could potentially have circular dependency when we
unload IB devices and also handle fib events:
1. The bond work starts first;
2. The "modprobe -rv bonding" process tries to release the bond device,
   with the "pernet_ops_rwsem" lock hold;
3. The bond work blocks in unregister_netdevice_notifier() and waits for
the lock because fib event came right before;
4. The kernel fib module tries to free all the fib entries by broadcasting
   the "FIB_EVENT_NH_DEL" event;
5. Upon the fib event this lag_mp module holds the fib lock and queue a
   fib work.
So:
   bond work -> modprobe task -> kernel fib module -> lag_mp -> bond work

Today we either reload IB devices in roce lag in nic mode or either handle
fib events in switchdev mode, but a new feature could change that we'll
need to reload IB devices also in switchdev mode so this is a future proof
fix as one may not notice this later.

Signed-off-by: Mark Zhang <markz@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agoMerge branch 'mlx5-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox...
Saeed Mahameed [Mon, 30 Mar 2020 06:41:50 +0000 (23:41 -0700)]
Merge branch 'mlx5-next' of git://git./linux/kernel/git/mellanox/linux

* 'mlx5-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux:
  mlx5: Remove uninitialized use of key in mlx5_core_create_mkey
  {IB,net}/mlx5: Move asynchronous mkey creation to mlx5_ib
  {IB,net}/mlx5: Assign mkey variant in mlx5_ib only
  {IB,net}/mlx5: Setup mkey variant before mr create command invocation

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agoMerge branch 'ethtool-netlink-interface-part-4'
David S. Miller [Mon, 30 Mar 2020 05:32:37 +0000 (22:32 -0700)]
Merge branch 'ethtool-netlink-interface-part-4'

Michal Kubecek says:

====================
ethtool netlink interface, part 4

Implementation of more netlink request types:

  - coalescing (ethtool -c/-C, patches 2-4)
  - pause parameters (ethtool -a/-A, patches 5-7)
  - EEE settings (--show-eee / --set-eee, patches 8-10)
  - timestamping info (-T, patches 11-12)

Patch 1 is a fix for netdev reference leak similar to commit 2f599ec422ad
("ethtool: fix reference leak in some *_SET handlers") but fixing a code

Changes in v3
  - change "one-step-*" Tx type names to "onestep-*", (patch 11, suggested
    by Richard Cochran
  - use "TSINFO" rather than "TIMESTAMP" for timestamping information
    constants and adjust symbol names (patch 12, suggested by Richard
    Cochran)

Changes in v2:
  - fix compiler warning in net_hwtstamp_validate() (patch 11)
  - fix follow-up lines alignment (whitespace only, patches 3 and 8)
which is only in net-next tree at the moment.
====================

Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: provide timestamping information with TSINFO_GET request
Michal Kubecek [Fri, 27 Mar 2020 23:01:58 +0000 (00:01 +0100)]
ethtool: provide timestamping information with TSINFO_GET request

Implement TSINFO_GET request to get timestamping information for a network
device. This is traditionally available via ETHTOOL_GET_TS_INFO ioctl
request.

Move part of ethtool_get_ts_info() into common.c so that ioctl and netlink
code use the same logic to get timestamping information from the device.

v3: use "TSINFO" rather than "TIMESTAMP", suggested by Richard Cochran

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: add timestamping related string sets
Michal Kubecek [Fri, 27 Mar 2020 23:01:53 +0000 (00:01 +0100)]
ethtool: add timestamping related string sets

Add three string sets related to timestamping information:

  ETH_SS_SOF_TIMESTAMPING: SOF_TIMESTAMPING_* flags
  ETH_SS_TS_TX_TYPES:      timestamping Tx types
  ETH_SS_TS_RX_FILTERS:    timestamping Rx filters

These will be used for TIMESTAMP_GET request.

v2: avoid compiler warning ("enumeration value not handled in switch")
    in net_hwtstamp_validate()

v3: omit dash in Tx type names ("one-step-*" -> "onestep-*"), suggested by
    Richard Cochran

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: add EEE_NTF notification
Michal Kubecek [Fri, 27 Mar 2020 23:01:48 +0000 (00:01 +0100)]
ethtool: add EEE_NTF notification

Send ETHTOOL_MSG_EEE_NTF notification whenever EEE settings of a network
device are modified using ETHTOOL_MSG_EEE_SET netlink message or
ETHTOOL_SEEE ioctl request.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: set EEE settings with EEE_SET request
Michal Kubecek [Fri, 27 Mar 2020 23:01:43 +0000 (00:01 +0100)]
ethtool: set EEE settings with EEE_SET request

Implement EEE_SET netlink request to set EEE settings of a network device.
These are traditionally set with ETHTOOL_SEEE ioctl request.

The netlink interface allows setting the EEE status for all link modes
supported by kernel but only first 32 link modes can be set at the moment
as only those are supported by the ethtool_ops callback.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: provide EEE settings with EEE_GET request
Michal Kubecek [Fri, 27 Mar 2020 23:01:38 +0000 (00:01 +0100)]
ethtool: provide EEE settings with EEE_GET request

Implement EEE_GET request to get EEE settings of a network device. These
are traditionally available via ETHTOOL_GEEE ioctl request.

The netlink interface allows reporting EEE status for all link modes
supported by kernel but only first 32 link modes are provided at the moment
as only those are reported by the ethtool_ops callback and drivers.

v2: fix alignment (whitespace only)

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: add PAUSE_NTF notification
Michal Kubecek [Fri, 27 Mar 2020 23:01:33 +0000 (00:01 +0100)]
ethtool: add PAUSE_NTF notification

Send ETHTOOL_MSG_PAUSE_NTF notification whenever pause parameters of
a network device are modified using ETHTOOL_MSG_PAUSE_SET netlink message
or ETHTOOL_SPAUSEPARAM ioctl request.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: set pause parameters with PAUSE_SET request
Michal Kubecek [Fri, 27 Mar 2020 23:01:28 +0000 (00:01 +0100)]
ethtool: set pause parameters with PAUSE_SET request

Implement PAUSE_SET netlink request to set pause parameters of a network
device. Thease are traditionally set with ETHTOOL_SPAUSEPARAM ioctl
request.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: provide pause parameters with PAUSE_GET request
Michal Kubecek [Fri, 27 Mar 2020 23:01:23 +0000 (00:01 +0100)]
ethtool: provide pause parameters with PAUSE_GET request

Implement PAUSE_GET request to get pause parameters of a network device.
These are traditionally available via ETHTOOL_GPAUSEPARAM ioctl request.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: add COALESCE_NTF notification
Michal Kubecek [Fri, 27 Mar 2020 23:01:18 +0000 (00:01 +0100)]
ethtool: add COALESCE_NTF notification

Send ETHTOOL_MSG_COALESCE_NTF notification whenever coalescing parameters
of a network device are modified using ETHTOOL_MSG_COALESCE_SET netlink
message or ETHTOOL_SCOALESCE ioctl request.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: set coalescing parameters with COALESCE_SET request
Michal Kubecek [Fri, 27 Mar 2020 23:01:13 +0000 (00:01 +0100)]
ethtool: set coalescing parameters with COALESCE_SET request

Implement COALESCE_SET netlink request to set coalescing parameters of
a network device. These are traditionally set with ETHTOOL_SCOALESCE ioctl
request. This commit adds only support for device coalescing parameters,
not per queue coalescing parameters.

Like the ioctl implementation, the generic ethtool code checks if only
supported parameters are modified; if not, first offending attribute is
reported using extack.

v2: fix alignment (whitespace only)

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: provide coalescing parameters with COALESCE_GET request
Michal Kubecek [Fri, 27 Mar 2020 23:01:08 +0000 (00:01 +0100)]
ethtool: provide coalescing parameters with COALESCE_GET request

Implement COALESCE_GET request to get coalescing parameters of a network
device. These are traditionally available via ETHTOOL_GCOALESCE ioctl
request. This commit adds only support for device coalescing parameters,
not per queue coalescing parameters.

Omit attributes with zero values unless they are declared as supported
(i.e. the corresponding bit in ethtool_ops::supported_coalesce_params is
set).

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: fix reference leak in ethnl_set_privflags()
Michal Kubecek [Fri, 27 Mar 2020 23:01:03 +0000 (00:01 +0100)]
ethtool: fix reference leak in ethnl_set_privflags()

Andrew noticed that some handlers for *_SET commands leak a netdev
reference if required ethtool_ops callbacks do not exist. One of them is
ethnl_set_privflags(), a simple reproducer would be e.g.

  ip link add veth1 type veth peer name veth2
  ethtool --set-priv-flags veth1 foo on
  ip link del veth1

Make sure dev_put() is called when ethtool_ops check fails.

Fixes: f265d799596a ("ethtool: set device private flags with PRIVFLAGS_SET request")
Reported-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'ipv6-add-rpl-source-routing'
David S. Miller [Mon, 30 Mar 2020 05:30:57 +0000 (22:30 -0700)]
Merge branch 'ipv6-add-rpl-source-routing'

Alexander Aring says:

====================
net: ipv6: add rpl source routing

This patch series will add handling for RPL source routing handling
and insertion (implement as lwtunnel)! I did an example prototype
implementation in rpld for using this implementation in non-storing mode:

https://github.com/linux-wpan/rpld/tree/nonstoring_mode

I will also present a talk at netdev about it:

https://netdevconf.info/0x14/session.html?talk-extend-segment-routing-for-RPL

In receive handling I add handling for IPIP encapsulation as RFC6554
describes it as possible. For reasons I didn't implemented it yet for
generating such packets because I am not really sure how/when this
should happen. So far I understand there exists a draft yet which
describes the cases (inclusive a Hop-by-Hop option which we also not
support yet).

https://tools.ietf.org/html/draft-ietf-roll-useofrplinfo-35

This is just the beginning to start implementation everything for yet,
step by step. It works for my use cases yet to have it running on a
6LOWPAN _only_ network.

I have some patches for iproute2 as well.

A sidenote: I check on local addresses if they are part of segment
routes, this is just to avoid stupid settings. A use can add addresses
afterwards what I cannot control anymore but then it's users fault to
make such thing. The receive handling checks for this as well which is
required by RFC6554, so the next hops or when it comes back should drop
it anyway.

To make this possible I added functionality to pass the net structure to
the build_state of lwtunnel (I hope I caught all lwtunnels).

Another sidenote: I set the headroom value to 0 as I figured out it will
break on interfaces with IPv6 min mtu if set to non zero for tunnels on
L3.

- Alex

changes since v3:
 - use parse_nested which isn't deprecated - Thanks David Ahern
 - change to return -1 instead errno in exthdr handling to unify
   error code
 - change function name from ipv6_rpl_srh_decompress_size to
   ipv6_rpl_srh_size

changes since v2:
 - add additional segdata length in lwtunnel build_state
 - fix build_state patch by not catching one inline noop function
   if LWTUNNEL is disabled

Alexander Aring (5):
  include: uapi: linux: add rpl sr header definition
  addrconf: add functionality to check on rpl requirements
  net: ipv6: add support for rpl sr exthdr
  net: add net available in build_state
  net: ipv6: add rpl sr tunnel
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: ipv6: add rpl sr tunnel
Alexander Aring [Fri, 27 Mar 2020 22:00:22 +0000 (18:00 -0400)]
net: ipv6: add rpl sr tunnel

This patch adds functionality to configure routes for RPL source routing
functionality. There is no IPIP functionality yet implemented which can
be added later when the cases when to use IPv6 encapuslation comes more
clear.

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: add net available in build_state
Alexander Aring [Fri, 27 Mar 2020 22:00:21 +0000 (18:00 -0400)]
net: add net available in build_state

The build_state callback of lwtunnel doesn't contain the net namespace
structure yet. This patch will add it so we can check on specific
address configuration at creation time of rpl source routes.

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: ipv6: add support for rpl sr exthdr
Alexander Aring [Fri, 27 Mar 2020 22:00:20 +0000 (18:00 -0400)]
net: ipv6: add support for rpl sr exthdr

This patch adds rpl source routing receive handling. Everything works
only if sysconf "rpl_seg_enabled" and source routing is enabled. Mostly
the same behaviour as IPv6 segmentation routing. To handle compression
and uncompression a rpl.c file is created which contains the necessary
functionality. The receive handling will also care about IPv6
encapsulated so far it's specified as possible nexthdr in RFC 6554.

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoaddrconf: add functionality to check on rpl requirements
Alexander Aring [Fri, 27 Mar 2020 22:00:19 +0000 (18:00 -0400)]
addrconf: add functionality to check on rpl requirements

This patch adds a functionality to addrconf to check on a specific RPL
address configuration. According to RFC 6554:

To detect loops in the SRH, a router MUST determine if the SRH
includes multiple addresses assigned to any interface on that
router. If such addresses appear more than once and are separated by
at least one address not assigned to that router.

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoinclude: uapi: linux: add rpl sr header definition
Alexander Aring [Fri, 27 Mar 2020 22:00:18 +0000 (18:00 -0400)]
include: uapi: linux: add rpl sr header definition

This patch adds a uapi header for rpl struct definition. The segments
data can be accessed over rpl_segaddr or rpl_segdata macros. In case of
compri and compre is zero the segment data is not compressed and can be
accessed by rpl_segaddr. In the other case the compressed data can be
accessed by rpl_segdata and interpreted as byte array.

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'mptcp-multiple-subflows-path-management'
David S. Miller [Mon, 30 Mar 2020 05:14:49 +0000 (22:14 -0700)]
Merge branch 'mptcp-multiple-subflows-path-management'

Mat Martineau says:

====================
Multipath TCP part 3: Multiple subflows and path management

v2 -> v3: Remove 'inline' in .c files, fix uapi bit macros, and rebase.

v1 -> v2: Rebase on current net-next, fix for netlink limit setting,
and update .gitignore for selftest.

This patch set allows more than one TCP subflow to be established and
used for a multipath TCP connection. Subflows are added to an existing
connection using the MP_JOIN option during the 3-way handshake. With
multiple TCP subflows available, sent data is now stored in the MPTCP
socket so it may be retransmitted on any TCP subflow if there is no
DATA_ACK before a timeout. If an MPTCP-level timeout occurs, data is
retransmitted using an available subflow. Storing this sent data
requires the addition of memory accounting at the MPTCP level, which was
previously delegated to the single subflow. Incoming DATA_ACKs now free
data from the MPTCP-level retransmit buffer.

IP addresses available for new subflow connections can now be advertised
and received with the ADD_ADDR option, and the corresponding REMOVE_ADDR
option likewise advertises that an address is no longer available.

The MPTCP path manager netlink interface has commands to set in-kernel
limits for the number of concurrent subflows and control the
advertisement of IP addresses between peers.

To track and debug MPTCP connections there are new MPTCP MIB counters,
and subflow context can be requested using inet_diag. The MPTCP
self-tests now validate multiple-subflow operation and the netlink path
manager interface.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoselftests: add test-cases for MPTCP MP_JOIN
Paolo Abeni [Fri, 27 Mar 2020 21:48:53 +0000 (14:48 -0700)]
selftests: add test-cases for MPTCP MP_JOIN

Use the pm netlink to configure the creation of several
subflows, and verify that via MIB counters.

Update the mptcp_connect program to allow reliable MP_JOIN
handshake even on small data file

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoselftests: add PM netlink functional tests
Paolo Abeni [Fri, 27 Mar 2020 21:48:52 +0000 (14:48 -0700)]
selftests: add PM netlink functional tests

This introduces basic self-tests for the PM netlink,
checking the basic APIs and possible exceptional
values.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: add netlink-based PM
Paolo Abeni [Fri, 27 Mar 2020 21:48:51 +0000 (14:48 -0700)]
mptcp: add netlink-based PM

Expose a new netlink family to userspace to control the PM, setting:

 - list of local addresses to be signalled.
 - list of local addresses used to created subflows.
 - maximum number of add_addr option to react

When the msk is fully established, the PM netlink attempts to
announce the 'signal' list via the ADD_ADDR option. Since we
currently lack the ADD_ADDR echo (and related event) only the
first addr is sent.

After exhausting the 'announce' list, the PM tries to create
subflow for each addr in 'local' list, waiting for each
connection to be completed before attempting the next one.

Idea is to add an additional PM hook for ADD_ADDR echo, to allow
the PM netlink announcing multiple addresses, in sequence.

Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: add and use MIB counter infrastructure
Florian Westphal [Fri, 27 Mar 2020 21:48:50 +0000 (14:48 -0700)]
mptcp: add and use MIB counter infrastructure

Exported via same /proc file as the Linux TCP MIB counters, so "netstat -s"
or "nstat" will show them automatically.

The MPTCP MIB counters are allocated in a distinct pcpu area in order to
avoid bloating/wasting TCP pcpu memory.

Counters are allocated once the first MPTCP socket is created in a
network namespace and free'd on exit.

If no sockets have been allocated, all-zero mptcp counters are shown.

The MIB counter list is taken from the multipath-tcp.org kernel, but
only a few counters have been picked up so far.  The counter list can
be increased at any time later on.

v2 -> v3:
 - remove 'inline' in foo.c files (David S. Miller)

Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: allow dumping subflow context to userspace
Davide Caratti [Fri, 27 Mar 2020 21:48:49 +0000 (14:48 -0700)]
mptcp: allow dumping subflow context to userspace

add ulp-specific diagnostic functions, so that subflow information can be
dumped to userspace programs like 'ss'.

v2 -> v3:
- uapi: use bit macros appropriate for userspace

Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: implement and use MPTCP-level retransmission
Paolo Abeni [Fri, 27 Mar 2020 21:48:48 +0000 (14:48 -0700)]
mptcp: implement and use MPTCP-level retransmission

On timeout event, schedule a work queue to do the retransmission.
Retransmission code closely resembles the sendmsg() implementation and
re-uses mptcp_sendmsg_frag, providing a dummy msghdr - for flags'
sake - and peeking the relevant dfrag from the rtx head.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: rework mptcp_sendmsg_frag to accept optional dfrag
Paolo Abeni [Fri, 27 Mar 2020 21:48:47 +0000 (14:48 -0700)]
mptcp: rework mptcp_sendmsg_frag to accept optional dfrag

This will simplify mptcp-level retransmission implementation
in the next patch. If dfrag is provided by the caller, skip
kernel space memory allocation and use data and metadata
provided by the dfrag itself.

Because a peer could ack data at TCP level but refrain from
sending mptcp-level ACKs, we could grow the mptcp socket
backlog indefinitely.

We should thus block mptcp_sendmsg until the peer has acked some of the
sent data.

In order to be able to do so, increment the mptcp socket wmem_queued
counter on memory allocation and decrement it when releasing the memory
on mptcp-level ack reception.

Because TCP performns sndbuf auto-tuning up to tcp_wmem_max[2], make
this the mptcp sk_sndbuf limit.

In the future we could add experiment with autotuning as TCP does in
tcp_sndbuf_expand().

v2 -> v3:
 - remove 'inline' in foo.c files (David S. Miller)

Co-developed-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: allow partial cleaning of rtx head dfrag
Florian Westphal [Fri, 27 Mar 2020 21:48:46 +0000 (14:48 -0700)]
mptcp: allow partial cleaning of rtx head dfrag

After adding wmem accounting for the mptcp socket we could get
into a situation where the mptcp socket can't transmit more data,
and mptcp_clean_una doesn't reduce wmem even if snd_una has advanced
because it currently will only remove entire dfrags.

Allow advancing the dfrag head sequence and reduce wmem,
even though this isn't correct (as we can't release the page).

Because we will soon block on mptcp sk in case wmem is too large,
call sk_stream_write_space() in case we reduced the backlog so
userspace task blocked in sendmsg or poll will be woken up.

This isn't an issue if the send buffer is large, but it is when
SO_SNDBUF is used to reduce it to a lower value.

Note we can still get a deadlock for low SO_SNDBUF values in
case both sides of the connection write to the socket: both could
be blocked due to wmem being too small -- and current mptcp stack
will only increment mptcp ack_seq on recv.

This doesn't happen with the selftest as it uses poll() and
will always call recv if there is data to read.

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: implement memory accounting for mptcp rtx queue
Paolo Abeni [Fri, 27 Mar 2020 21:48:45 +0000 (14:48 -0700)]
mptcp: implement memory accounting for mptcp rtx queue

Charge the data on the rtx queue to the master MPTCP socket, too.
Such memory in uncharged when the data is acked/dequeued.

Also account mptcp sockets inuse via a protocol specific pcpu
counter.

Co-developed-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: introduce MPTCP retransmission timer
Paolo Abeni [Fri, 27 Mar 2020 21:48:44 +0000 (14:48 -0700)]
mptcp: introduce MPTCP retransmission timer

The timer will be used to schedule retransmission. It's
frequency is based on the current subflow RTO estimation and
is reset on every una_seq update

The timer is clearer for good by __mptcp_clear_xmit()

Also clean MPTCP rtx queue before each transmission.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: queue data for mptcp level retransmission
Paolo Abeni [Fri, 27 Mar 2020 21:48:43 +0000 (14:48 -0700)]
mptcp: queue data for mptcp level retransmission

Keep the send page fragment on an MPTCP level retransmission queue.
The queue entries are allocated inside the page frag allocator,
acquiring an additional reference to the page for each list entry.

Also switch to a custom page frag refill function, to ensure that
the current page fragment can always host an MPTCP rtx queue entry.

The MPTCP rtx queue is flushed at disconnect() and close() time

Note that now we need to call __mptcp_init_sock() regardless of mptcp
enable status, as the destructor will try to walk the rtx_queue.

v2 -> v3:
 - remove 'inline' in foo.c files (David S. Miller)

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: update per unacked sequence on pkt reception
Paolo Abeni [Fri, 27 Mar 2020 21:48:42 +0000 (14:48 -0700)]
mptcp: update per unacked sequence on pkt reception

So that we keep per unacked sequence number consistent; since
we update per msk data, use an atomic64 cmpxchg() to protect
against concurrent updates from multiple subflows.

Initialize the snd_una at connect()/accept() time.

Co-developed-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: Implement path manager interface commands
Peter Krystad [Fri, 27 Mar 2020 21:48:41 +0000 (14:48 -0700)]
mptcp: Implement path manager interface commands

Fill in more path manager functionality by adding a worker function and
modifying the related stub functions to schedule the worker.

Co-developed-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Peter Krystad <peter.krystad@linux.intel.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: Add handling of outgoing MP_JOIN requests
Peter Krystad [Fri, 27 Mar 2020 21:48:40 +0000 (14:48 -0700)]
mptcp: Add handling of outgoing MP_JOIN requests

Subflow creation may be initiated by the path manager when
the primary connection is fully established and a remote
address has been received via ADD_ADDR.

Create an in-kernel sock and use kernel_connect() to
initiate connection.

Passive sockets can't acquire the mptcp socket lock at
subflow creation time, so an additional list protected by
a new spinlock is used to track the MPJ subflows.

Such list is spliced into conn_list tail every time the msk
socket lock is acquired, so that it will not interfere
with data flow on the original connection.

Data flow and connection failover not addressed by this commit.

Co-developed-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Peter Krystad <peter.krystad@linux.intel.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: Add handling of incoming MP_JOIN requests
Peter Krystad [Fri, 27 Mar 2020 21:48:39 +0000 (14:48 -0700)]
mptcp: Add handling of incoming MP_JOIN requests

Process the MP_JOIN option in a SYN packet with the same flow
as MP_CAPABLE but when the third ACK is received add the
subflow to the MPTCP socket subflow list instead of adding it to
the TCP socket accept queue.

The subflow is added at the end of the subflow list so it will not
interfere with the existing subflows operation and no data is
expected to be transmitted on it.

Co-developed-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Peter Krystad <peter.krystad@linux.intel.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: Add path manager interface
Peter Krystad [Fri, 27 Mar 2020 21:48:38 +0000 (14:48 -0700)]
mptcp: Add path manager interface

Add enough of a path manager interface to allow sending of ADD_ADDR
when an incoming MPTCP connection is created. Capable of sending only
a single IPv4 ADD_ADDR option. The 'pm_data' element of the connection
sock will need to be expanded to handle multiple interfaces and IPv6.
Partial processing of the incoming ADD_ADDR is included so the path
manager notification of that event happens at the proper time, which
involves validating the incoming address information.

This is a skeleton interface definition for events generated by
MPTCP.

Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Co-developed-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Peter Krystad <peter.krystad@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomptcp: Add ADD_ADDR handling
Peter Krystad [Fri, 27 Mar 2020 21:48:37 +0000 (14:48 -0700)]
mptcp: Add ADD_ADDR handling

Add handling for sending and receiving the ADD_ADDR, ADD_ADDR6,
and RM_ADDR suboptions.

Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Peter Krystad <peter.krystad@linux.intel.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlx4: fix "initializer element not constant" compiler error
Jacob Keller [Fri, 27 Mar 2020 21:08:35 +0000 (14:08 -0700)]
mlx4: fix "initializer element not constant" compiler error

A recent commit e8937681797c ("devlink: prepare to support region
operations") used the region_cr_space_str and region_fw_health_str
variables as initializers for the devlink_region_ops structures.

This can result in compiler errors:
drivers/net/ethernet/mellanox//mlx4/crdump.c:45:10: error: initializer
element is not constant
   .name = region_cr_space_str,
           ^
drivers/net/ethernet/mellanox//mlx4/crdump.c:45:10: note: (near
initialization for ‘region_cr_space_ops.name’)
drivers/net/ethernet/mellanox//mlx4/crdump.c:50:10: error: initializer
element is not constant
   .name = region_fw_health_str,

The variables were made to be "const char * const", indicating that both
the pointer and data were constant. This was enough to resolve this on
recent GCC (gcc (GCC) 9.2.1 20190827 (Red Hat 9.2.1-1) for this author).

Unfortunately this is not enough for older compilers to realize that the
variable can be treated as a constant expression.

Fix this by introducing macros for the string and use those instead of
the variable name in the region ops structures.

Reported-by: tanhuazhong <tanhuazhong@huawei.com>
Fixes: e8937681797c ("devlink: prepare to support region operations")
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agodevlink: don't wrap commands in rST shell blocks
Jacob Keller [Fri, 27 Mar 2020 20:55:36 +0000 (13:55 -0700)]
devlink: don't wrap commands in rST shell blocks

The devlink-region.rst and ice-region.rst documentation files wrapped
some lines within shell code blocks due to being longer than 80 lines.

It was pointed out during review that wrapping these lines shouldn't be
done. Fix these two rST files and remove the line wrapping on these
shell command examples.

Reported-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: dsa: mt7530: use resolved link config in mac_link_up()
René van Dorst [Fri, 27 Mar 2020 14:44:12 +0000 (15:44 +0100)]
net: dsa: mt7530: use resolved link config in mac_link_up()

Convert the mt7530 switch driver to use the finalised link
parameters in mac_link_up() rather than the parameters in mac_config().

Signed-off-by: René van Dorst <opensource@vdorst.com>
Tested-by: Sean Wang <sean.wang@mediatek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: dsa: sja1105: show more ethtool statistics counters for P/Q/R/S
Vladimir Oltean [Fri, 27 Mar 2020 14:00:16 +0000 (16:00 +0200)]
net: dsa: sja1105: show more ethtool statistics counters for P/Q/R/S

It looks like the P/Q/R/S series supports some more counters,
generically named "Ethernet statistics counter", which we were not
printing. Add them.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agos390/qeth: support net namespaces for L3 devices
Julian Wiedmann [Fri, 27 Mar 2020 11:00:42 +0000 (12:00 +0100)]
s390/qeth: support net namespaces for L3 devices

Enable the L3 driver's IPv4 address notifier to watch for events on qeth
devices that have been moved into a net namespace. We need to program
those IPs into the HW just as usual, otherwise inbound traffic won't
flow.

Fixes: 6133fb1aa137 ("[NETNS]: Disable inetaddr notifiers in namespaces other than initial.")
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: Fix typo of SKB_SGO_CB_OFFSET
Cambda Zhu [Thu, 26 Mar 2020 07:33:14 +0000 (15:33 +0800)]
net: Fix typo of SKB_SGO_CB_OFFSET

The SKB_SGO_CB_OFFSET should be SKB_GSO_CB_OFFSET which means the
offset of the GSO in skb cb. This patch fixes the typo.

Fixes: 9207f9d45b0a ("net: preserve IP control block during GSO segmentation")
Signed-off-by: Cambda Zhu <cambda@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: Fix race condition between scheduling and destroying the slowpath workqueue
Yuval Basson [Wed, 25 Mar 2020 20:50:43 +0000 (22:50 +0200)]
qed: Fix race condition between scheduling and destroying the slowpath workqueue

Calling queue_delayed_work concurrently with
destroy_workqueue might race to an unexpected outcome -
scheduled task after wq is destroyed or other resources
(like ptt_pool) are freed (yields NULL pointer dereference).
cancel_delayed_work prevents the race by cancelling
the timer triggered for scheduling a new task.

Fixes: 59ccf86fe ("qed: Add driver infrastucture for handling mfw requests")
Signed-off-by: Denis Bolotin <dbolotin@marvell.com>
Signed-off-by: Michal Kalderon <mkalderon@marvell.com>
Signed-off-by: Yuval Basson <ybason@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: page pool: allow to pass zero flags to page_pool_init()
Denis Kirjanov [Wed, 25 Mar 2020 20:35:28 +0000 (23:35 +0300)]
net: page pool: allow to pass zero flags to page_pool_init()

page pool API can be useful for non-DMA cases like
xen-netfront driver so let's allow to pass zero flags to
page pool flags.

v2: check DMA direction only if PP_FLAG_DMA_MAP is set

Signed-off-by: Denis Kirjanov <kda@linux-powerpc.org>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoselftests: move timestamping selftests to net folder
Jian Yang [Wed, 25 Mar 2020 20:32:07 +0000 (13:32 -0700)]
selftests: move timestamping selftests to net folder

For historical reasons, there are several timestamping selftest targets
in selftests/networking/timestamping. Move them to the standard
directory for networking tests: selftests/net.

Signed-off-by: Jian Yang <jianyang@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoARM: dts: apalis-imx6qdl: use rgmii-id instead of rgmii
Philippe Schenker [Wed, 25 Mar 2020 17:34:25 +0000 (18:34 +0100)]
ARM: dts: apalis-imx6qdl: use rgmii-id instead of rgmii

Until now a PHY-fixup in mach-imx set our rgmii timing correctly. For
the PHY KSZ9131 there is no PHY-fixup in mach-imx. To support this PHY
too, use rgmii-id.
For the now used KSZ9031 nothing will change, as rgmii-id is only
implemented and supported by the KSZ9131.

Signed-off-by: Philippe Schenker <philippe.schenker@toradex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: phy: micrel.c: add rgmii interface delay possibility to ksz9131
Philippe Schenker [Wed, 25 Mar 2020 17:34:24 +0000 (18:34 +0100)]
net: phy: micrel.c: add rgmii interface delay possibility to ksz9131

The KSZ9131 provides DLL controlled delays on RXC and TXC lines. This
patch makes use of those delays. The information which delays should
be enabled or disabled comes from the interface names, documented in
ethernet-controller.yaml:

rgmii:      Disable RXC and TXC delays
rgmii-id:   Enable RXC and TXC delays
rgmii-txid: Enable only TXC delay, disable RXC delay
rgmii-rxid: Enable onlx RXC delay, disable TXC delay

Signed-off-by: Philippe Schenker <philippe.schenker@toradex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: macsec: add support for specifying offload upon link creation
Mark Starovoytov [Wed, 25 Mar 2020 13:01:34 +0000 (16:01 +0300)]
net: macsec: add support for specifying offload upon link creation

This patch adds new netlink attribute to allow a user to (optionally)
specify the desired offload mode immediately upon MACSec link creation.

Separate iproute patch will be required to support this from user space.

Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
David S. Miller [Mon, 30 Mar 2020 04:25:29 +0000 (21:25 -0700)]
Merge git://git./linux/kernel/git/davem/net

Minor comment conflict in mac80211.

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoLinux 5.6
Linus Torvalds [Sun, 29 Mar 2020 22:25:41 +0000 (15:25 -0700)]
Linux 5.6

4 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Sun, 29 Mar 2020 17:40:31 +0000 (10:40 -0700)]
Merge branch 'akpm' (patches from Andrew)

Merge vm fixes from Andrew Morton:
 "5 fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  mm/sparse: fix kernel crash with pfn_section_valid check
  mm: fork: fix kernel_stack memcg stats for various stack implementations
  hugetlb_cgroup: fix illegal access to memory
  drivers/base/memory.c: indicate all memory blocks as removable
  mm/swapfile.c: move inode_lock out of claim_swapfile

4 years agoMerge tag 'timers-urgent-2020-03-29' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 29 Mar 2020 17:36:29 +0000 (10:36 -0700)]
Merge tag 'timers-urgent-2020-03-29' of git://git./linux/kernel/git/tip/tip

Pull timer fix from Thomas Gleixner:
 "A single fix for the Hyper-V clocksource driver to make sched clock
  actually return nanoseconds and not the virtual clock value which
  increments at 10e7 HZ (100ns)"

* tag 'timers-urgent-2020-03-29' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  clocksource/drivers/hyper-v: Make sched clock return nanoseconds correctly

4 years agoMerge tag 'irq-urgent-2020-03-29' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 29 Mar 2020 17:07:00 +0000 (10:07 -0700)]
Merge tag 'irq-urgent-2020-03-29' of git://git./linux/kernel/git/tip/tip

Pull irq fix from Thomas Gleixner:
 "A single bugfix to prevent reference leaks in irq affinity notifiers"

* tag 'irq-urgent-2020-03-29' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  genirq: Fix reference leaks on irq affinity notifiers

4 years agomm/sparse: fix kernel crash with pfn_section_valid check
Aneesh Kumar K.V [Sun, 29 Mar 2020 02:17:29 +0000 (19:17 -0700)]
mm/sparse: fix kernel crash with pfn_section_valid check

Fix the crash like this:

    BUG: Kernel NULL pointer dereference on read at 0x00000000
    Faulting instruction address: 0xc000000000c3447c
    Oops: Kernel access of bad area, sig: 11 [#1]
    LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
    CPU: 11 PID: 7519 Comm: lt-ndctl Not tainted 5.6.0-rc7-autotest #1
    ...
    NIP [c000000000c3447c] vmemmap_populated+0x98/0xc0
    LR [c000000000088354] vmemmap_free+0x144/0x320
    Call Trace:
       section_deactivate+0x220/0x240
       __remove_pages+0x118/0x170
       arch_remove_memory+0x3c/0x150
       memunmap_pages+0x1cc/0x2f0
       devm_action_release+0x30/0x50
       release_nodes+0x2f8/0x3e0
       device_release_driver_internal+0x168/0x270
       unbind_store+0x130/0x170
       drv_attr_store+0x44/0x60
       sysfs_kf_write+0x68/0x80
       kernfs_fop_write+0x100/0x290
       __vfs_write+0x3c/0x70
       vfs_write+0xcc/0x240
       ksys_write+0x7c/0x140
       system_call+0x5c/0x68

The crash is due to NULL dereference at

test_bit(idx, ms->usage->subsection_map);

due to ms->usage = NULL in pfn_section_valid()

With commit d41e2f3bd546 ("mm/hotplug: fix hot remove failure in
SPARSEMEM|!VMEMMAP case") section_mem_map is set to NULL after
depopulate_section_mem().  This was done so that pfn_page() can work
correctly with kernel config that disables SPARSEMEM_VMEMMAP.  With that
config pfn_to_page does

__section_mem_map_addr(__sec) + __pfn;

where

  static inline struct page *__section_mem_map_addr(struct mem_section *section)
  {
unsigned long map = section->section_mem_map;
map &= SECTION_MAP_MASK;
return (struct page *)map;
  }

Now with SPASEMEM_VMEMAP enabled, mem_section->usage->subsection_map is
used to check the pfn validity (pfn_valid()).  Since section_deactivate
release mem_section->usage if a section is fully deactivated,
pfn_valid() check after a subsection_deactivate cause a kernel crash.

  static inline int pfn_valid(unsigned long pfn)
  {
  ...
return early_section(ms) || pfn_section_valid(ms, pfn);
  }

where

  static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
  {
int idx = subsection_map_index(pfn);

return test_bit(idx, ms->usage->subsection_map);
  }

Avoid this by clearing SECTION_HAS_MEM_MAP when mem_section->usage is
freed.  For architectures like ppc64 where large pages are used for
vmmemap mapping (16MB), a specific vmemmap mapping can cover multiple
sections.  Hence before a vmemmap mapping page can be freed, the kernel
needs to make sure there are no valid sections within that mapping.
Clearing the section valid bit before depopulate_section_memap enables
this.

[aneesh.kumar@linux.ibm.com: add comment]
Link: http://lkml.kernel.org/r/20200326133235.343616-1-aneesh.kumar@linux.ibm.comLink:
Fixes: d41e2f3bd546 ("mm/hotplug: fix hot remove failure in SPARSEMEM|!VMEMMAP case")
Reported-by: Sachin Sant <sachinp@linux.vnet.ibm.com>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Tested-by: Sachin Sant <sachinp@linux.vnet.ibm.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4 years agomm: fork: fix kernel_stack memcg stats for various stack implementations
Roman Gushchin [Sun, 29 Mar 2020 02:17:25 +0000 (19:17 -0700)]
mm: fork: fix kernel_stack memcg stats for various stack implementations

Depending on CONFIG_VMAP_STACK and the THREAD_SIZE / PAGE_SIZE ratio the
space for task stacks can be allocated using __vmalloc_node_range(),
alloc_pages_node() and kmem_cache_alloc_node().

In the first and the second cases page->mem_cgroup pointer is set, but
in the third it's not: memcg membership of a slab page should be
determined using the memcg_from_slab_page() function, which looks at
page->slab_cache->memcg_params.memcg .  In this case, using
mod_memcg_page_state() (as in account_kernel_stack()) is incorrect:
page->mem_cgroup pointer is NULL even for pages charged to a non-root
memory cgroup.

It can lead to kernel_stack per-memcg counters permanently showing 0 on
some architectures (depending on the configuration).

In order to fix it, let's introduce a mod_memcg_obj_state() helper,
which takes a pointer to a kernel object as a first argument, uses
mem_cgroup_from_obj() to get a RCU-protected memcg pointer and calls
mod_memcg_state().  It allows to handle all possible configurations
(CONFIG_VMAP_STACK and various THREAD_SIZE/PAGE_SIZE values) without
spilling any memcg/kmem specifics into fork.c .

Note: This is a special version of the patch created for stable
backports.  It contains code from the following two patches:
  - mm: memcg/slab: introduce mem_cgroup_from_obj()
  - mm: fork: fix kernel_stack memcg stats for various stack implementations

[guro@fb.com: introduce mem_cgroup_from_obj()]
Link: http://lkml.kernel.org/r/20200324004221.GA36662@carbon.dhcp.thefacebook.com
Fixes: 4d96ba353075 ("mm: memcg/slab: stop setting page->mem_cgroup pointer for slab pages")
Signed-off-by: Roman Gushchin <guro@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Bharata B Rao <bharata@linux.ibm.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: <stable@vger.kernel.org>
Link: http://lkml.kernel.org/r/20200303233550.251375-1-guro@fb.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4 years agohugetlb_cgroup: fix illegal access to memory
Mina Almasry [Sun, 29 Mar 2020 02:17:22 +0000 (19:17 -0700)]
hugetlb_cgroup: fix illegal access to memory

This appears to be a mistake in commit faced7e0806cf ("mm: hugetlb
controller for cgroups v2").

Essentially that commit does a hugetlb_cgroup_from_counter assuming that
page_counter_try_charge has initialized counter.

But if that has failed then it seems will not initialize counter, so
hugetlb_cgroup_from_counter(counter) ends up pointing to random memory,
causing kasan to complain.

The solution is to simply use 'h_cg', instead of
hugetlb_cgroup_from_counter(counter), since that is a reference to the
hugetlb_cgroup anyway.  After this change kasan ceases to complain.

Fixes: faced7e0806cf ("mm: hugetlb controller for cgroups v2")
Reported-by: syzbot+cac0c4e204952cf449b1@syzkaller.appspotmail.com
Signed-off-by: Mina Almasry <almasrymina@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Giuseppe Scrivano <gscrivan@redhat.com>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Link: http://lkml.kernel.org/r/20200313223920.124230-1-almasrymina@google.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4 years agodrivers/base/memory.c: indicate all memory blocks as removable
David Hildenbrand [Sun, 29 Mar 2020 02:17:19 +0000 (19:17 -0700)]
drivers/base/memory.c: indicate all memory blocks as removable

We see multiple issues with the implementation/interface to compute
whether a memory block can be offlined (exposed via
/sys/devices/system/memory/memoryX/removable) and would like to simplify
it (remove the implementation).

1. It runs basically lockless. While this might be good for performance,
   we see possible races with memory offlining that will require at
   least some sort of locking to fix.

2. Nowadays, more false positives are possible. No arch-specific checks
   are performed that validate if memory offlining will not be denied
   right away (and such check will require locking). For example, arm64
   won't allow to offline any memory block that was added during boot -
   which will imply a very high error rate. Other archs have other
   constraints.

3. The interface is inherently racy. E.g., if a memory block is detected
   to be removable (and was not a false positive at that time), there is
   still no guarantee that offlining will actually succeed. So any
   caller already has to deal with false positives.

4. It is unclear which performance benefit this interface actually
   provides. The introducing commit 5c755e9fd813 ("memory-hotplug: add
   sysfs removable attribute for hotplug memory remove") mentioned

"A user-level agent must be able to identify which sections
 of memory are likely to be removable before attempting the
 potentially expensive operation."

   However, no actual performance comparison was included.

Known users:

 - lsmem: Will group memory blocks based on the "removable" property. [1]

 - chmem: Indirect user. It has a RANGE mode where one can specify
          removable ranges identified via lsmem to be offlined. However,
          it also has a "SIZE" mode, which allows a sysadmin to skip the
          manual "identify removable blocks" step. [2]

 - powerpc-utils: Uses the "removable" attribute to skip some memory
          blocks right away when trying to find some to offline+remove.
          However, with ballooning enabled, it already skips this
          information completely (because it once resulted in many false
          negatives). Therefore, the implementation can deal with false
          positives properly already. [3]

According to Nathan Fontenot, DLPAR on powerpc is nowadays no longer
driven from userspace via the drmgr command (powerpc-utils).  Nowadays
it's managed in the kernel - including onlining/offlining of memory
blocks - triggered by drmgr writing to /sys/kernel/dlpar.  So the
affected legacy userspace handling is only active on old kernels.  Only
very old versions of drmgr on a new kernel (unlikely) might execute
slower - totally acceptable.

With CONFIG_MEMORY_HOTREMOVE, always indicating "removable" should not
break any user space tool.  We implement a very bad heuristic now.
Without CONFIG_MEMORY_HOTREMOVE we cannot offline anything, so report
"not removable" as before.

Original discussion can be found in [4] ("[PATCH RFC v1] mm:
is_mem_section_removable() overhaul").

Other users of is_mem_section_removable() will be removed next, so that
we can remove is_mem_section_removable() completely.

[1] http://man7.org/linux/man-pages/man1/lsmem.1.html
[2] http://man7.org/linux/man-pages/man8/chmem.8.html
[3] https://github.com/ibm-power-utilities/powerpc-utils
[4] https://lkml.kernel.org/r/20200117105759.27905-1-david@redhat.com

Also, this patch probably fixes a crash reported by Steve.
http://lkml.kernel.org/r/CAPcyv4jpdaNvJ67SkjyUJLBnBnXXQv686BiVW042g03FUmWLXw@mail.gmail.com

Reported-by: "Scargall, Steve" <steve.scargall@intel.com>
Suggested-by: Michal Hocko <mhocko@kernel.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Nathan Fontenot <ndfont@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Badari Pulavarty <pbadari@us.ibm.com>
Cc: Robert Jennings <rcj@linux.vnet.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Karel Zak <kzak@redhat.com>
Cc: <stable@vger.kernel.org>
Link: http://lkml.kernel.org/r/20200128093542.6908-1-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>