openwrt/staging/blogic.git
5 years agoice: Update rings based on TC information
Anirudh Venkataramanan [Thu, 28 Feb 2019 23:24:27 +0000 (15:24 -0800)]
ice: Update rings based on TC information

This patch adds a new function ice_vsi_cfg_dcb_rings which updates a
VSI's rings based on DCB traffic class information.

Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoice: Add code to process LLDP MIB change events
Anirudh Venkataramanan [Thu, 28 Feb 2019 23:24:26 +0000 (15:24 -0800)]
ice: Add code to process LLDP MIB change events

This patch adds support to process LLDP MIB change notifications sent
by the firmware.

Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoice: Add code for DCB initialization part 4/4
Anirudh Venkataramanan [Thu, 28 Feb 2019 23:24:25 +0000 (15:24 -0800)]
ice: Add code for DCB initialization part 4/4

When the firmware doesn't support LLDP or DCBX, the driver should switch
to "software LLDP mode". This patch adds support for the same.

Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoice: Add code for DCB initialization part 3/4
Anirudh Venkataramanan [Thu, 28 Feb 2019 23:24:24 +0000 (15:24 -0800)]
ice: Add code for DCB initialization part 3/4

This patch adds a new function ice_pf_dcb_cfg (and related helpers)
which applies the DCB configuration obtained from the firmware. As
part of this, VSIs/netdevs are updated with traffic class information.

This patch requires a bit of a refactor of existing code.

1. For a MIB change event, the associated VSI is closed and brought up
   again. The gap between closing and opening the VSI can cause a race
   condition. Fix this by grabbing the rtnl_lock prior to closing the
   VSI and then only free it after re-opening the VSI during a MIB
   change event.

2. ice_sched_query_elem is used in ice_sched.c and with this patch, in
   ice_dcb.c as well. However, ice_dcb.c is not built when CONFIG_DCB is
   unset. This results in namespace warnings (ice_sched.o: Externally
   defined symbols with no external references) when CONFIG_DCB is unset.
   To avoid this move ice_sched_query_elem from ice_sched.c to
   ice_common.c.

Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoice: Add code for DCB initialization part 2/4
Anirudh Venkataramanan [Thu, 28 Feb 2019 23:24:23 +0000 (15:24 -0800)]
ice: Add code for DCB initialization part 2/4

This patch introduces a new top level function ice_init_dcb (and
related lower level helper functions) which continues the DCB init
flow.

This function uses ice_get_dcb_cfg to get, parse and store the DCB
configuration. Once this is done, it sets itself up to be notified
by the firmware on LLDP MIB change events.

Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoice: Add code for DCB initialization part 1/4
Anirudh Venkataramanan [Thu, 28 Feb 2019 23:24:22 +0000 (15:24 -0800)]
ice: Add code for DCB initialization part 1/4

This patch introduces a skeleton for ice_init_pf_dcb, the top level
function for DCB initialization. Subsequent patches will add to this
DCB init flow.

In this patch, ice_init_pf_dcb checks if DCB is a supported capability.
If so, an admin queue call to start the LLDP and DCBx in firmware is
issued. If not, an error is reported. Note that we don't fail the driver
init if DCB init fails.

Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoice: Bump version
Anirudh Venkataramanan [Tue, 19 Feb 2019 23:04:14 +0000 (15:04 -0800)]
ice: Bump version

Bump driver version to 0.7.3

Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoice: Fix incorrect use of abbreviations
Anirudh Venkataramanan [Tue, 19 Feb 2019 23:04:13 +0000 (15:04 -0800)]
ice: Fix incorrect use of abbreviations

Capitalize abbreviations and spell out some that aren't obvious.

Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoice: Fix typos in code comments
Anirudh Venkataramanan [Tue, 19 Feb 2019 23:04:12 +0000 (15:04 -0800)]
ice: Fix typos in code comments

This patch fixes typos in code comments.

Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoMerge branch 'ipv6-Use-fib6_result-for-fib_lookups'
David S. Miller [Thu, 18 Apr 2019 06:11:52 +0000 (23:11 -0700)]
Merge branch 'ipv6-Use-fib6_result-for-fib_lookups'

David Ahern says:

====================
ipv6: Use fib6_result for fib_lookups

Add fib6_result as a single data structure to hold results from a fib
lookup. IPv6 currently has everything in 1 data structure - a fib6_info,
but with nexthop objects the fib6_nh can be in a nexthop or a nexthop
can be a blackhole which affects the fib6_type and flags (REJECT).

v2
- fixed 2 bugs in patch12:
  i. checking return from fib6_table_lookup in fib6_lookup
  ii. call to fib6_rule_saddr in fib6_rule_action_alt should use res->nh
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Add fib6_type and fib6_flags to fib6_result
David Ahern [Tue, 16 Apr 2019 21:36:11 +0000 (14:36 -0700)]
ipv6: Add fib6_type and fib6_flags to fib6_result

Add the fib6_flags and fib6_type to fib6_result. Update the lookup helpers
to set them and update post fib lookup users to use the version from the
result.

This allows nexthop objects to have blackhole nexthop.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to fib lookups
David Ahern [Tue, 16 Apr 2019 21:36:10 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to fib lookups

Change fib6_lookup and fib6_table_lookup to take a fib6_result and set
f6i and nh rather than returning a fib6_info. For now both always
return 0.

A later patch set can make these more like the IPv4 counterparts and
return EINVAL, EACCESS, etc based on fib6_type.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to fib6_table_lookup tracepoint
David Ahern [Tue, 16 Apr 2019 21:36:09 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to fib6_table_lookup tracepoint

Change fib6_table_lookup tracepoint to take the fib6_result and use
the fib6_info and fib6_nh from it.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to rt6_select and find_rr_leaf
David Ahern [Tue, 16 Apr 2019 21:36:08 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to rt6_select and find_rr_leaf

Pass fib6_result to rt6_select. Instead of returning the fib entry, it
will set f6i and nh based on the lookup.

find_rr_leaf is changed to remove the match option in favor of taking
fib6_result and having __find_rr_leaf set f6i in the result.

In the process, update fib6_info references in __find_rr_leaf to f6i names.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to rt6_device_match
David Ahern [Tue, 16 Apr 2019 21:36:07 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to rt6_device_match

Pass fib6_result to rt6_device_match with f6i set. rt6_device_match
updates f6i in the result if it finds a better match and sets nh.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to ip6_mtu_from_fib6 and fib6_mtu
David Ahern [Tue, 16 Apr 2019 21:36:06 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to ip6_mtu_from_fib6 and fib6_mtu

Change ip6_mtu_from_fib6 and fib6_mtu to take a fib6_result over a
fib6_info. Update both to use the fib6_nh from fib6_result.

Since the signature of ip6_mtu_from_fib6 is already changing, add const
to daddr and saddr.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to rt6_insert_exception
David Ahern [Tue, 16 Apr 2019 21:36:05 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to rt6_insert_exception

Update rt6_insert_exception to take a fib6_result over a fib6_info.
Change ort to f6i from the fib6_result and rename to better reflect
what it references (a fib6_info).

Since this function is already getting changed, update the comments
to reference fib6_info variables rather than the older rt6_info.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to ip6_rt_get_dev_rcu and ip6_rt_copy_init
David Ahern [Tue, 16 Apr 2019 21:36:04 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to ip6_rt_get_dev_rcu and ip6_rt_copy_init

Now that all callers are update to have a fib6_result, pass it down
to ip6_rt_get_dev_rcu, ip6_rt_copy_init, and ip6_rt_init_dst.

In the process, change ort to f6i in ip6_rt_copy_init to make it
clear it is a reference to a fib6_info.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to pcpu route functions
David Ahern [Tue, 16 Apr 2019 21:36:03 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to pcpu route functions

Update ip6_rt_pcpu_alloc, rt6_get_pcpu_route and rt6_make_pcpu_route
to a fib6_result over a fib6_info.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to ip6_create_rt_rcu
David Ahern [Tue, 16 Apr 2019 21:36:02 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to ip6_create_rt_rcu

Change ip6_create_rt_rcu to take fib6_result over a fib6_info.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to ip6_rt_cache_alloc
David Ahern [Tue, 16 Apr 2019 21:36:01 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to ip6_rt_cache_alloc

Change ip6_rt_cache_alloc to take a fib6_result over a fib6_info.

Since ip6_rt_cache_alloc is only the caller, update the
rt6_is_gw_or_nonexthop helper to take fib6_result.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Pass fib6_result to rt6_find_cached_rt
David Ahern [Tue, 16 Apr 2019 21:36:00 +0000 (14:36 -0700)]
ipv6: Pass fib6_result to rt6_find_cached_rt

Simplify rt6_find_cached_rt for the fast path cases and pass fib6_result
to rt6_find_cached_rt. Rename the local return variable to ret to maintain
consisting with fib6_result name.

Update the comment in rt6_find_cached_rt to reference the new names in
a fib6_info vs the old name when fib entries were an rt6_info.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Rename fib6_multipath_select and pass fib6_result
David Ahern [Tue, 16 Apr 2019 21:35:59 +0000 (14:35 -0700)]
ipv6: Rename fib6_multipath_select and pass fib6_result

Add 'struct fib6_result' to hold the fib entry and fib6_nh from a fib
lookup as separate entries, similar to what IPv4 now has with fib_result.

Rename fib6_multipath_select to fib6_select_path, pass fib6_result to
it, and set f6i and nh in the result once a path selection is done.
Call fib6_select_path unconditionally for path selection which means
moving the sibling and oif check to fib6_select_path. To handle the two
different call paths (2 only call multipath_select if flowi6_oif == 0 and
the other always calls it), add a new have_oif_match that controls the
sibling walk if relevant.

Update callers of fib6_multipath_select accordingly and have them use the
fib6_info and fib6_nh from the result.

This is needed for multipath nexthop objects where a single f6i can
point to multiple fib6_nh (similar to IPv4).

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
David S. Miller [Wed, 17 Apr 2019 18:26:25 +0000 (11:26 -0700)]
Merge git://git./linux/kernel/git/davem/net

Conflict resolution of af_smc.c from Stephen Rothwell.

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 's390-next'
David S. Miller [Wed, 17 Apr 2019 17:33:59 +0000 (10:33 -0700)]
Merge branch 's390-next'

Julian Wiedmann says:

====================
s390/qeth: updates 2019-04-17

please apply some additional qeth patches to net-next. This patchset
converts the driver to use the kernel's multiqueue model.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agos390/qeth: stop/wake TX queues based on their fill level
Julian Wiedmann [Wed, 17 Apr 2019 16:17:35 +0000 (18:17 +0200)]
s390/qeth: stop/wake TX queues based on their fill level

Current xmit code only stops the txq after attempting to fill an
IO buffer that hasn't been TX-completed yet. In many-connection
scenarios, this can result in frequent rejected TX attempts, requeuing
of skbs with NETDEV_TX_BUSY and extra overhead.

Now that we have a proper 1-to-1 relation between stack-side txqs and
our HW Queues, overhaul the stop/wake logic so that the xmit code
stops the txq as needed.
Given that we might map multiple skbs into a single buffer, it's crucial
to ensure that the queue always provides an _entirely_ empty IO buffer.
Otherwise large skbs (eg TSO) might not fit into the last available
buffer. So whenever qeth_do_send_packet() first utilizes an _empty_
buffer, it updates & checks the used_buffers count.

This now ensures that an skb passed to qeth_xmit() can always be mapped
into an IO buffer, so remove all of the -EBUSY roll-back handling in the
TX path. We preserve the minimal safety-checks ("Is this IO buffer
really available?"), just in case some nasty future bug ever attempts to
corrupt an in-use buffer.

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agos390/qeth: simplify QoS code
Julian Wiedmann [Wed, 17 Apr 2019 16:17:34 +0000 (18:17 +0200)]
s390/qeth: simplify QoS code

qeth_get_priority_queue() is no longer used for IQD devices, remove the
special-casing of their mcast queue.

This effectively reverts
commit 70deb01662b1 ("qeth: omit outbound queue 3 for unicast packets in Priority Queuing on HiperSockets").

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agos390/qeth: add TX multiqueue support for OSA devices
Julian Wiedmann [Wed, 17 Apr 2019 16:17:33 +0000 (18:17 +0200)]
s390/qeth: add TX multiqueue support for OSA devices

This adds trivial support for multiple TX queues on OSA-style devices
(both real HW and z/VM NICs). For now we expose the driver's existing
QoS mechanism via .ndo_select_queue, and adjust the number of available
TX queues when qeth_update_from_chp_desc() detects that the
HW configuration has changed.

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agos390/qeth: add TX multiqueue support for IQD devices
Julian Wiedmann [Wed, 17 Apr 2019 16:17:32 +0000 (18:17 +0200)]
s390/qeth: add TX multiqueue support for IQD devices

qeth has been supporting multiple HW Output Queues for a long time. But
rather than exposing those queues to the stack, it uses its own queue
selection logic in .ndo_start_xmit... with all the drawbacks that
entails.
Start off by switching IQD devices over to a proper mqs net_device,
and converting all the netdev_queue management code.

One oddity with IQD devices is the requirement to place all mcast
traffic on the _highest_ established HW queue. Doing so via
.ndo_select_queue seems straight-forward - but that won't work if only
some of the HW queues are active
(ie. when dev->real_num_tx_queues < dev->num_tx_queues), since
netdev_cap_txqueue() will not allow us to put skbs on the higher queues.

To make this work, we
1. let .ndo_select_queue() map all mcast traffic to netdev_queue 0, and
2. later re-map the netdev_queue and HW queue indices in
   .ndo_start_xmit and the TX completion handler.

With this patch we default to a fixed set of 1 ucast and 1 mcast queue.
Support for dynamic reconfiguration is added at a later time.

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agos390/qeth: don't keep statistics for tx timeout
Julian Wiedmann [Wed, 17 Apr 2019 16:17:31 +0000 (18:17 +0200)]
s390/qeth: don't keep statistics for tx timeout

struct netdev_queue contains a counter for tx timeouts, which gets
updated by dev_watchdog(). So let's not attempt to maintain our own
statistics, in particular not by overloading the skb-error counter.

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agos390/qeth: don't bother updating the last-tx time
Julian Wiedmann [Wed, 17 Apr 2019 16:17:30 +0000 (18:17 +0200)]
s390/qeth: don't bother updating the last-tx time

As the documentation for netif_trans_update() says, netdev_start_xmit()
already updates the last-tx time after every good xmit. So don't
duplicate that effort.

One odd case is that qeth_flush_buffers() also gets called from our
TX completion handler, to flush out any partially filled buffer when
we switch the queue to non-packing mode. But as the TX completion
handler will _always_ wake the txq, we don't have to worry about
the TX watchdog there.

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agos390/qeth: handle error from qeth_update_from_chp_desc()
Julian Wiedmann [Wed, 17 Apr 2019 16:17:29 +0000 (18:17 +0200)]
s390/qeth: handle error from qeth_update_from_chp_desc()

Subsequent code relies on the values that qeth_update_from_chp_desc()
reads from the CHP descriptor. Rather than dealing with weird errors
later on, just handle it properly here.

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agos390/qeth: clarify naming for some QDIO helpers
Julian Wiedmann [Wed, 17 Apr 2019 16:17:28 +0000 (18:17 +0200)]
s390/qeth: clarify naming for some QDIO helpers

The naming of several QDIO helpers doesn't match their actual
functionality, or the structures they operate on. Clean this up.

s/qeth_alloc_qdio_buffers/qeth_alloc_qdio_queues
s/qeth_free_qdio_buffers/qeth_free_qdio_queues
s/qeth_alloc_qdio_out_buf/qeth_alloc_output_queue
s/qeth_clear_outq_buffers/qeth_drain_output_queue
s/qeth_clear_qdio_buffers/qeth_drain_output_queues

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next...
David S. Miller [Wed, 17 Apr 2019 17:31:21 +0000 (10:31 -0700)]
Merge branch '40GbE' of git://git./linux/kernel/git/jkirsher/next-queue

Jeff Kirsher says:

====================
40GbE Intel Wired LAN Driver Updates 2019-04-16

This series contains updates to i40e driver only.

Adam fixes i40e so that queues can be restored to its original value if
configuring queue channels fails.  Bumped the maximum API version
supported and added the API version to error messages to clarify
supported firmware API versions.  Fixed the problem with the driver
being able to add only 7 multicast MAC address filters instead of 16.

Aleksandr adds support for Dynamic Device Personalization (DDP) which
allows loading profiles that change the way internal parser interprets
processed frames.

Nick fixes an issue where if we modify the VLAN stripping options when a
port VLAN is configured, it will break traffic for the VSI, so prevent
changes from being made.

Jake fixes an issue where a device reset can mess up the clock time
because we reset the clock time based on the kernel time every reset.
This causes us to potentially completely reset the PTP time, and can
cause unexpected behavior in programs like ptp4l.

Piotr fixes an LED blink issue with the 'ethtool -p' command, so that
identification blinking will work on all hardware.

Chinh fixed the error returned to correctly reflect the current state
when LLDP or DCBx is not in an operational state.

Grzegorz cleans up a misleading error message when untrusted VF tries to
exceed addresses beyond the NIC limit.

Carolyn fixes the error return code to correctly reflect the error case.

v2: updated the URL provided in the DDP patch (#2)
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge tag 'for-linus-5.1-2' of git://github.com/cminyard/linux-ipmi
Linus Torvalds [Wed, 17 Apr 2019 17:25:25 +0000 (10:25 -0700)]
Merge tag 'for-linus-5.1-2' of git://github.com/cminyard/linux-ipmi

Pull IPMI fixes from Corey Minyard:
 "Fixes for some bugs cause by recent changes. One crash if you feed bad
  data to the module parameters, one BUG that sometimes occurs when a
  user closes the connection, and one bug that cause the driver to not
  work if the configuration information only comes in from SMBIOS"

* tag 'for-linus-5.1-2' of git://github.com/cminyard/linux-ipmi:
  ipmi: fix sleep-in-atomic in free_user at cleanup SRCU user->release_barrier
  ipmi: ipmi_si_hardcode.c: init si_type array to fix a crash
  ipmi: Fix failure on SMBIOS specified devices

5 years agoMerge branch 'stmmac-Enable-Flow-Control'
David S. Miller [Wed, 17 Apr 2019 17:14:28 +0000 (10:14 -0700)]
Merge branch 'stmmac-Enable-Flow-Control'

Jose Abreu says:

====================
net: stmmac: Enable Flow Control

I don't know of any specific reason why Flow Control is off by default but
do let me know if there is any.

Tested in B2B between XGMAC2 and GMAC5.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: stmmac: Set Flow Control to automatic mode in the driver
Jose Abreu [Wed, 17 Apr 2019 07:33:05 +0000 (09:33 +0200)]
net: stmmac: Set Flow Control to automatic mode in the driver

By default Flow Control feature is not being enabled in stmmac.

This is a useful feature that can prevent loss of packets and now that
XGMAC already supports it (along with GMAC and QoS) it makes sense to
activate it.

Switch the module parameter to FLOW_AUTO so that Flow Control is
activated.

Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: stmmac: dwxgmac: Finish the Flow Control implementation
Jose Abreu [Wed, 17 Apr 2019 07:33:04 +0000 (09:33 +0200)]
net: stmmac: dwxgmac: Finish the Flow Control implementation

Finish the implementation of Flow Control feature. In order for it to
work correctly we need to set EHFC bit and the correct threshold values
for activating and deactivating it.

Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Wed, 17 Apr 2019 16:57:45 +0000 (09:57 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Handle init flow failures properly in iwlwifi driver, from Shahar S
    Matityahu.

 2) mac80211 TXQs need to be unscheduled on powersave start, from Felix
    Fietkau.

 3) SKB memory accounting fix in A-MDSU aggregation, from Felix Fietkau.

 4) Increase RCU lock hold time in mlx5 FPGA code, from Saeed Mahameed.

 5) Avoid checksum complete with XDP in mlx5, also from Saeed.

 6) Fix netdev feature clobbering in ibmvnic driver, from Thomas Falcon.

 7) Partial sent TLS record leak fix from Jakub Kicinski.

 8) Reject zero size iova range in vhost, from Jason Wang.

 9) Allow pending work to complete before clcsock release from Karsten
    Graul.

10) Fix XDP handling max MTU in thunderx, from Matteo Croce.

11) A lot of protocols look at the sa_family field of a sockaddr before
    validating it's length is large enough, from Tetsuo Handa.

12) Don't write to free'd pointer in qede ptp error path, from Colin Ian
    King.

13) Have to recompile IP options in ipv4_link_failure because it can be
    invoked from ARP, from Stephen Suryaputra.

14) Doorbell handling fixes in qed from Denis Bolotin.

15) Revert net-sysfs kobject register leak fix, it causes new problems.
    From Wang Hai.

16) Spectre v1 fix in ATM code, from Gustavo A. R. Silva.

17) Fix put of BROPT_VLAN_STATS_PER_PORT in bridging code, from Nikolay
    Aleksandrov.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (111 commits)
  socket: fix compat SO_RCVTIMEO_NEW/SO_SNDTIMEO_NEW
  tcp: tcp_grow_window() needs to respect tcp_space()
  ocelot: Clean up stats update deferred work
  ocelot: Don't sleep in atomic context (irqs_disabled())
  net: bridge: fix netlink export of vlan_stats_per_port option
  qed: fix spelling mistake "faspath" -> "fastpath"
  tipc: set sysctl_tipc_rmem and named_timeout right range
  tipc: fix link established but not in session
  net: Fix missing meta data in skb with vlan packet
  net: atm: Fix potential Spectre v1 vulnerabilities
  net/core: work around section mismatch warning for ptp_classifier
  net: bridge: fix per-port af_packet sockets
  bnx2x: fix spelling mistake "dicline" -> "decline"
  route: Avoid crash from dereferencing NULL rt->from
  MAINTAINERS: normalize Woojung Huh's email address
  bonding: fix event handling for stacked bonds
  Revert "net-sysfs: Fix memory leak in netdev_register_kobject"
  rtnetlink: fix rtnl_valid_stats_req() nlmsg_len check
  qed: Fix the DORQ's attentions handling
  qed: Fix missing DORQ attentions
  ...

5 years agoipmi: fix sleep-in-atomic in free_user at cleanup SRCU user->release_barrier
Corey Minyard [Wed, 3 Apr 2019 20:58:16 +0000 (15:58 -0500)]
ipmi: fix sleep-in-atomic in free_user at cleanup SRCU user->release_barrier

free_user() could be called in atomic context.

This patch pushed the free operation off into a workqueue.

Example:

 BUG: sleeping function called from invalid context at kernel/workqueue.c:2856
 in_atomic(): 1, irqs_disabled(): 0, pid: 177, name: ksoftirqd/27
 CPU: 27 PID: 177 Comm: ksoftirqd/27 Not tainted 4.19.25-3 #1
 Hardware name: AIC 1S-HV26-08/MB-DPSB04-06, BIOS IVYBV060 10/21/2015
 Call Trace:
  dump_stack+0x5c/0x7b
  ___might_sleep+0xec/0x110
  __flush_work+0x48/0x1f0
  ? try_to_del_timer_sync+0x4d/0x80
  _cleanup_srcu_struct+0x104/0x140
  free_user+0x18/0x30 [ipmi_msghandler]
  ipmi_free_recv_msg+0x3a/0x50 [ipmi_msghandler]
  deliver_response+0xbd/0xd0 [ipmi_msghandler]
  deliver_local_response+0xe/0x30 [ipmi_msghandler]
  handle_one_recv_msg+0x163/0xc80 [ipmi_msghandler]
  ? dequeue_entity+0xa0/0x960
  handle_new_recv_msgs+0x15c/0x1f0 [ipmi_msghandler]
  tasklet_action_common.isra.22+0x103/0x120
  __do_softirq+0xf8/0x2d7
  run_ksoftirqd+0x26/0x50
  smpboot_thread_fn+0x11d/0x1e0
  kthread+0x103/0x140
  ? sort_range+0x20/0x20
  ? kthread_destroy_worker+0x40/0x40
  ret_from_fork+0x1f/0x40

Fixes: 77f8269606bf ("ipmi: fix use-after-free of user->release_barrier.rda")
Reported-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: stable@vger.kernel.org # 5.0
Cc: Yang Yingliang <yangyingliang@huawei.com>
5 years agosocket: fix compat SO_RCVTIMEO_NEW/SO_SNDTIMEO_NEW
Arnd Bergmann [Tue, 16 Apr 2019 20:31:14 +0000 (22:31 +0200)]
socket: fix compat SO_RCVTIMEO_NEW/SO_SNDTIMEO_NEW

It looks like the new socket options only work correctly
for native execution, but in case of compat mode fall back
to the old behavior as we ignore the 'old_timeval' flag.

Rework so we treat SO_RCVTIMEO_NEW/SO_SNDTIMEO_NEW the
same way in compat and native 32-bit mode.

Cc: Deepa Dinamani <deepa.kernel@gmail.com>
Fixes: a9beb86ae6e5 ("sock: Add SO_RCVTIMEO_NEW and SO_SNDTIMEO_NEW")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Deepa Dinamani <deepa.kernel@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agotcp: tcp_grow_window() needs to respect tcp_space()
Eric Dumazet [Tue, 16 Apr 2019 17:55:20 +0000 (10:55 -0700)]
tcp: tcp_grow_window() needs to respect tcp_space()

For some reason, tcp_grow_window() correctly tests if enough room
is present before attempting to increase tp->rcv_ssthresh,
but does not prevent it to grow past tcp_space()

This is causing hard to debug issues, like failing
the (__tcp_select_window(sk) >= tp->rcv_wnd) test
in __tcp_ack_snd_check(), causing ACK delays and possibly
slow flows.

Depending on tcp_rmem[2], MTU, skb->len/skb->truesize ratio,
we can see the problem happening on "netperf -t TCP_RR -- -r 2000,2000"
after about 60 round trips, when the active side no longer sends
immediate acks.

This bug predates git history.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Wei Wang <weiwan@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'dpaa2-eth-Add-flow-steering-support-without-masking'
David S. Miller [Wed, 17 Apr 2019 04:46:19 +0000 (21:46 -0700)]
Merge branch 'dpaa2-eth-Add-flow-steering-support-without-masking'

Ioana Ciocoi Radulescu says:

====================
dpaa2-eth: Add flow steering support without masking

On DPAA2 platforms that lack a TCAM (like LS1088A), masking of
flow steering keys is not supported. Until now we didn't offer
flow steering capabilities at all on these platforms.

Introduce a limited support for flow steering, where we only
allow ethtool rules that share a common key (i.e. have the same
header fields). If a rule with a new composition key is wanted,
the user must first manually delete all previous rules.

First patch fixes a minor bug, the next two cleanup and prepare
the code and the last one introduces the actual FS support.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agodpaa2-eth: Add flow steering support without masking
Ioana Ciocoi Radulescu [Tue, 16 Apr 2019 17:13:30 +0000 (17:13 +0000)]
dpaa2-eth: Add flow steering support without masking

On platforms that lack a TCAM (like LS1088A), masking of
flow steering keys is not supported. Until now we didn't
offer flow steering capabilities at all on these platforms,
since our driver implementation configured a "comprehensive"
FS key (containing all supported header fields), with masks
used to ignore the fields not present in the rules provided
by the user.

We now allow ethtool rules that share a common key (i.e. have
the same header fields). The FS key is now kept in the driver
private data and initialized when the first rule is added to
an empty table, rather than at probe time. If a rule with a new
composition key is wanted, the user must first manually delete
all previous rules.

When building a FS table entry to pass to firmware, we still use
the old building algorithm, which assumes an all-supported-fields
key, and later collapse the fields which aren't actually needed.

Masked rules are not supported; if provided, the mask value
will be ignored. For firmware versions older than MC10.7.0
(that only offer the legacy ABIs for configuring distribution
keys) flow steering without masking support remains unavailable.

Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agodpaa2-eth: Update hash key composition code
Ioana Ciocoi Radulescu [Tue, 16 Apr 2019 17:13:29 +0000 (17:13 +0000)]
dpaa2-eth: Update hash key composition code

Introduce an internal id bitfield to uniquely identify header fields
supported by the Rx distribution keys. For the hash key, add a
conversion from the RXH_* bitmask provided by ethtool to the internal
ids.

Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agodpaa2-eth: Add a couple of macros
Ioana Ciocoi Radulescu [Tue, 16 Apr 2019 17:13:29 +0000 (17:13 +0000)]
dpaa2-eth: Add a couple of macros

Add two macros to simplify reading DPNI options.

Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agodpaa2-eth: Fix Rx classification status
Ioana Ciocoi Radulescu [Tue, 16 Apr 2019 17:13:28 +0000 (17:13 +0000)]
dpaa2-eth: Fix Rx classification status

Set the Rx flow classification enable flag only if key config
operation is successful.

Fixes 3f9b5c9 ("dpaa2-eth: Configure Rx flow classification key")

Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoocelot: Clean up stats update deferred work
Claudiu Manoil [Tue, 16 Apr 2019 14:51:59 +0000 (17:51 +0300)]
ocelot: Clean up stats update deferred work

This is preventive cleanup that may save troubles later.
No need to cancel repeateadly queued work if code is properly
refactored.
Don't let the ethtool -s process interfere with the stat workqueue
scheduling.

Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoocelot: Don't sleep in atomic context (irqs_disabled())
Claudiu Manoil [Tue, 16 Apr 2019 14:51:58 +0000 (17:51 +0300)]
ocelot: Don't sleep in atomic context (irqs_disabled())

Preemption disabled at:
 [<ffff000008cabd54>] dev_set_rx_mode+0x1c/0x38
 Call trace:
 [<ffff00000808a5c0>] dump_backtrace+0x0/0x3d0
 [<ffff00000808a9a4>] show_stack+0x14/0x20
 [<ffff000008e6c0c0>] dump_stack+0xac/0xe4
 [<ffff0000080fe76c>] ___might_sleep+0x164/0x238
 [<ffff0000080fe890>] __might_sleep+0x50/0x88
 [<ffff0000082261e4>] kmem_cache_alloc+0x17c/0x1d0
 [<ffff000000ea0ae8>] ocelot_set_rx_mode+0x108/0x188 [mscc_ocelot_common]
 [<ffff000008cabcf0>] __dev_set_rx_mode+0x58/0xa0
 [<ffff000008cabd5c>] dev_set_rx_mode+0x24/0x38

Fixes: a556c76adc05 ("net: mscc: Add initial Ocelot switch support")
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonfp: flower: fix implicit fallthrough warning
John Hurley [Tue, 16 Apr 2019 14:04:23 +0000 (15:04 +0100)]
nfp: flower: fix implicit fallthrough warning

The nfp_flower_copy_pre_actions function introduces a case statement with
an intentional fallthrough. However, this generates a warning if built
with the -Wimplicit-fallthrough flag.

Remove the warning by adding a fall through comment.

Fixes: 1c6952ca587d ("nfp: flower: generate merge flow rule")
Signed-off-by: John Hurley <john.hurley@netronome.com>
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Reviewed-by: Jiong Wang <jiong.wang@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: bridge: fix netlink export of vlan_stats_per_port option
Nikolay Aleksandrov [Tue, 16 Apr 2019 13:15:56 +0000 (16:15 +0300)]
net: bridge: fix netlink export of vlan_stats_per_port option

Since the introduction of the vlan_stats_per_port option the netlink
export of it has been broken since I made a typo and used the ifla
attribute instead of the bridge option to retrieve its state.
Sysfs export is fine, only netlink export has been affected.

Fixes: 9163a0fc1f0c0 ("net: bridge: add support for per-port vlan stats")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoqed: fix spelling mistake "faspath" -> "fastpath"
Colin Ian King [Tue, 16 Apr 2019 11:43:17 +0000 (12:43 +0100)]
qed: fix spelling mistake "faspath" -> "fastpath"

There is a spelling mistake in a DP_INFO message, fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: phy: micrel: add Asym Pause workaround
Antoine Tenart [Tue, 16 Apr 2019 10:10:20 +0000 (12:10 +0200)]
net: phy: micrel: add Asym Pause workaround

The Micrel KSZ9031 PHY may fail to establish a link when the Asymmetric
Pause capability is set. This issue is described in a Silicon Errata
(DS80000691D or DS80000692D), which advises to always disable the
capability. This patch implements the workaround by defining a KSZ9031
specific get_feature callback to force the Asymmetric Pause capability
bit to be cleared.

This fixes issues where the link would not come up at boot time, or when
the Asym Pause bit was set later on.

Signed-off-by: Antoine Tenart <antoine.tenart@bootlin.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'bnx2x-Support-for-timestamping-in-P2P-mode'
David S. Miller [Wed, 17 Apr 2019 04:34:48 +0000 (21:34 -0700)]
Merge branch 'bnx2x-Support-for-timestamping-in-P2P-mode'

Sudarsana Reddy Kalluru says:

====================
bnx2x: Support for timestamping in P2P mode.

The patch series adds driver support for timestamping the ptp packets
in peer-delay (P2P) mode.
 - Patch (1) performs the code cleanup.
 - Patch (2) adds the required implementation.

Please consider applying it 'net-next' tree.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnx2x: Add support for detection of P2P event packets.
Sudarsana Reddy Kalluru [Tue, 16 Apr 2019 08:46:13 +0000 (01:46 -0700)]
bnx2x: Add support for detection of P2P event packets.

The patch adds support for detecting the P2P (peer-to-peer) event packets.
This is required for timestamping the PTP packets in peer delay mode.
Unmask the below bits (set to 0) for device to detect the p2p packets.
  NIG_REG_P0/1_LLH_PTP_PARAM_MASK
  NIG_REG_P0/1_TLLH_PTP_PARAM_MASK
    bit 1 - IPv4 DA 1 of 224.0.0.107.
    bit 3 - IPv6 DA 1 of 0xFF02:0:0:0:0:0:0:6B.
    bit 9 - MAC DA 1 of 0x01-80-C2-00-00-0E.
  NIG_REG_P0/1_LLH_PTP_RULE_MASK
  NIG_REG_P0/1_TLLH_PTP_RULE_MASK
    bit 2 - {IPv4 DA 1; UDP DP 0}
    bit 6 - MAC Ethertype 0 of 0x88F7.
    bit 9 - MAC DA 1 of 0x01-80-C2-00-00-0E.

Signed-off-by: Sudarsana Reddy Kalluru <skalluru@marvell.com>
Signed-off-by: Ariel Elior <aelior@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnx2x: Replace magic numbers with macro definitions.
Sudarsana Reddy Kalluru [Tue, 16 Apr 2019 08:46:12 +0000 (01:46 -0700)]
bnx2x: Replace magic numbers with macro definitions.

This patch performs code cleanup by defining macros for the ptp-timestamp
filters.

Signed-off-by: Sudarsana Reddy Kalluru <skalluru@marvell.com>
Signed-off-by: Ariel Elior <aelior@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agotipc: set sysctl_tipc_rmem and named_timeout right range
Jie Liu [Tue, 16 Apr 2019 05:10:09 +0000 (13:10 +0800)]
tipc: set sysctl_tipc_rmem and named_timeout right range

We find that sysctl_tipc_rmem and named_timeout do not have the right minimum
setting. sysctl_tipc_rmem should be larger than zero, like sysctl_tcp_rmem.
And named_timeout as a timeout setting should be not less than zero.

Fixes: cc79dd1ba9c10 ("tipc: change socket buffer overflow control to respect sk_rcvbuf")
Fixes: a5325ae5b8bff ("tipc: add name distributor resiliency queue")
Signed-off-by: Jie Liu <liujie165@huawei.com>
Reported-by: Qiang Ning <ningqiang1@huawei.com>
Reviewed-by: Zhiqiang Liu <liuzhiqiang26@huawei.com>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agotipc: fix link established but not in session
Tuong Lien [Tue, 16 Apr 2019 03:48:07 +0000 (10:48 +0700)]
tipc: fix link established but not in session

According to the link FSM, when a link endpoint got RESET_MSG (- a
traditional one without the stopping bit) from its peer, it moves to
PEER_RESET state and raises a LINK_DOWN event which then resets the
link itself. Its state will become ESTABLISHING after the reset event
and the link will be re-established soon after this endpoint starts to
send ACTIVATE_MSG to the peer.

There is no problem with this mechanism, however the link resetting has
cleared the link 'in_session' flag (along with the other important link
data such as: the link 'mtu') that was correctly set up at the 1st step
(i.e. when this endpoint received the peer RESET_MSG). As a result, the
link will become ESTABLISHED, but the 'in_session' flag is not set, and
all STATE_MSG from its peer will be dropped at the link_validate_msg().
It means the link not synced and will sooner or later face a failure.

Since the link reset action is obviously needed for a new link session
(this is also true in the other situations), the problem here is that
the link is re-established a bit too early when the link endpoints are
not really in-sync yet. The commit forces a resync as already done in
the previous commit 91986ee166cf ("tipc: fix link session and
re-establish issues") by simply varying the link 'peer_session' value
at the link_reset().

Acked-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Tuong Lien <tuong.t.lien@dektech.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: Fix missing meta data in skb with vlan packet
Yuya Kusakabe [Tue, 16 Apr 2019 01:22:28 +0000 (10:22 +0900)]
net: Fix missing meta data in skb with vlan packet

skb_reorder_vlan_header() should move XDP meta data with ethernet header
if XDP meta data exists.

Fixes: de8f3a83b0a0 ("bpf: add meta pointer for direct access")
Signed-off-by: Yuya Kusakabe <yuya.kusakabe@gmail.com>
Signed-off-by: Takeru Hayasaka <taketarou2@gmail.com>
Co-developed-by: Takeru Hayasaka <taketarou2@gmail.com>
Reviewed-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoxen-netfront: mark expected switch fall-through
Gustavo A. R. Silva [Mon, 15 Apr 2019 21:11:41 +0000 (16:11 -0500)]
xen-netfront: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch
cases where we are expecting to fall through.

This patch fixes the following warning:

drivers/net/xen-netfront.c: In function ‘netback_changed’:
drivers/net/xen-netfront.c:2038:6: warning: this statement may fall through [-Wimplicit-fallthrough=]
   if (dev->state == XenbusStateClosed)
      ^
drivers/net/xen-netfront.c:2041:2: note: here
  case XenbusStateClosing:
  ^~~~

Warning level 3 was used: -Wimplicit-fallthrough=3

Notice that, in this particular case, the code comment is modified
in accordance with what GCC is expecting to find.

This patch is part of the ongoing efforts to enable
-Wimplicit-fallthrough.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: atm: Fix potential Spectre v1 vulnerabilities
Gustavo A. R. Silva [Mon, 15 Apr 2019 20:57:23 +0000 (15:57 -0500)]
net: atm: Fix potential Spectre v1 vulnerabilities

arg is controlled by user-space, hence leading to a potential
exploitation of the Spectre variant 1 vulnerability.

This issue was detected with the help of Smatch:

net/atm/lec.c:715 lec_mcast_attach() warn: potential spectre issue 'dev_lec' [r] (local cap)

Fix this by sanitizing arg before using it to index dev_lec.

Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].

[1] https://lore.kernel.org/lkml/20180423164740.GY17484@dhcp22.suse.cz/

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet/core: work around section mismatch warning for ptp_classifier
Ard Biesheuvel [Mon, 15 Apr 2019 17:14:45 +0000 (19:14 +0200)]
net/core: work around section mismatch warning for ptp_classifier

The routine ptp_classifier_init() uses an initializer for an
automatic struct type variable which refers to an __initdata
symbol. This is perfectly legal, but may trigger a section
mismatch warning when running the compiler in -fpic mode, due
to the fact that the initializer may be emitted into an anonymous
.data section thats lack the __init annotation. So work around it
by using assignments instead.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: bridge: fix per-port af_packet sockets
Nikolay Aleksandrov [Thu, 11 Apr 2019 10:56:39 +0000 (13:56 +0300)]
net: bridge: fix per-port af_packet sockets

When the commit below was introduced it changed two visible things:
 - the skb was no longer passed through the protocol handlers with the
   original device
 - the skb was passed up the stack with skb->dev = bridge

The first change broke af_packet sockets on bridge ports. For example we
use them for hostapd which listens for ETH_P_PAE packets on the ports.
We discussed two possible fixes:
 - create a clone and pass it through NF_HOOK(), act on the original skb
   based on the result
 - somehow signal to the caller from the okfn() that it was called,
   meaning the skb is ok to be passed, which this patch is trying to
   implement via returning 1 from the bridge link-local okfn()

Note that we rely on the fact that NF_QUEUE/STOLEN would return 0 and
drop/error would return < 0 thus the okfn() is called only when the
return was 1, so we signal to the caller that it was called by preserving
the return value from nf_hook().

Fixes: 8626c56c8279 ("bridge: fix potential use-after-free when hook returns QUEUE or STOLEN verdict")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoi40e: Fix misleading error message
Carolyn Wyborny [Wed, 6 Feb 2019 23:08:26 +0000 (15:08 -0800)]
i40e: Fix misleading error message

This patch changes an error code for an admin queue
head overrun to use I40E_ERR_ADMIN_QUEUE_FULL instead
of I40E_ERR_QUEUE_EMPTY.

Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: Able to add up to 16 MAC filters on an untrusted VF
Adam Ludkiewicz [Wed, 6 Feb 2019 23:08:25 +0000 (15:08 -0800)]
i40e: Able to add up to 16 MAC filters on an untrusted VF

This patch fixes the problem with the driver being able to add only 7
multicast MAC address filters instead of 16. The problem is fixed by
changing the maximum number of MAC address filters to 16+1+1 (two extra
are needed because the driver uses 1 for unicast MAC address and 1 for
broadcast).

Signed-off-by: Adam Ludkiewicz <adam.ludkiewicz@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: Report advertised link modes on 40GBASE_SR4
Adam Ludkiewicz [Wed, 6 Feb 2019 23:08:24 +0000 (15:08 -0800)]
i40e: Report advertised link modes on 40GBASE_SR4

Defined the advertised link mode field for 40000baseSR4_Full for
use with ethtool.

Signed-off-by: Adam Ludkiewicz <adam.ludkiewicz@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: The driver now prints the API version in error message
Adam Ludkiewicz [Wed, 6 Feb 2019 23:08:23 +0000 (15:08 -0800)]
i40e: The driver now prints the API version in error message

Added the API version in the error message for clarity.

Signed-off-by: Adam Ludkiewicz <adam.ludkiewicz@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: Changed maximum supported FW API version to 1.8
Adam Ludkiewicz [Wed, 6 Feb 2019 23:08:22 +0000 (15:08 -0800)]
i40e: Changed maximum supported FW API version to 1.8

A new FW has been released, which uses API version 1.8.

Signed-off-by: Adam Ludkiewicz <adam.ludkiewicz@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: Remove misleading messages for untrusted VF
Grzegorz Siwik [Wed, 6 Feb 2019 23:08:21 +0000 (15:08 -0800)]
i40e: Remove misleading messages for untrusted VF

Removed misleading messages when untrusted VF tries to
add more addresses than NIC limit

Signed-off-by: Grzegorz Siwik <grzegorz.siwik@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: Update i40e_init_dcb to return correct error
Chinh T Cao [Wed, 6 Feb 2019 23:08:20 +0000 (15:08 -0800)]
i40e: Update i40e_init_dcb to return correct error

Modify the i40e_init_dcb to return the correct error when LLDP or DCBX
is not in operational state.

Signed-off-by: Chinh T Cao <chinh.t.cao@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: Fix for 10G ports LED not blinking
Piotr Marczak [Wed, 6 Feb 2019 23:08:19 +0000 (15:08 -0800)]
i40e: Fix for 10G ports LED not blinking

On some hardware LEDs would not blink after command 'ethtool -p {eth-port}'
in certain circumstances. Now, function does not care about the activity
of the LED (though still preserves its state) but forcibly executes
identification blinking and then restores the LED state.

Signed-off-by: Piotr Marczak <piotr.marczak@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: save PTP time before a device reset
Jacob Keller [Tue, 12 Feb 2019 21:56:24 +0000 (13:56 -0800)]
i40e: save PTP time before a device reset

In the case where PTP is running on the hardware clock, but the kernel
system time is not being synced, a device reset can mess up the clock
time.

This occurs because we reset the clock time based on the kernel time
every reset. This causes us to potentially completely reset the PTP
time, and can cause unexpected behavior in programs like ptp4l.

Avoid this by saving the PTP time prior to device reset, and then
restoring using that time after the reset.

Directly restoring the PTP time we saved isn't perfect, because time
should have continued running, but the clock will essentially be stopped
during the reset. This is still better than the current solution of
assuming that the PTP HW clock is synced to the CLOCK_REALTIME.

We can do even better, by saving the ktime and calculating
a differential, using ktime_get(). This is based on CLOCK_MONOTONIC, and
allows us to get a fairly precise measure of the time difference between
saving and restoring the time.

Using this, we can update the saved PTP time, and use that as the value
to write to the hardware clock registers. This, of course is not perfect.
However, it does help ensure that the PTP time is restored as close as
feasible to the time it should have been if the reset had not occurred.

During device initialization, continue using the system time as the
source for the creation of the PTP clock, since this is the best known
current time source at driver load.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: don't allow changes to HW VLAN stripping on active port VLANs
Nicholas Nunley [Wed, 6 Feb 2019 23:08:17 +0000 (15:08 -0800)]
i40e: don't allow changes to HW VLAN stripping on active port VLANs

Modifying the VLAN stripping options when a port VLAN is configured
will break traffic for the VSI, and conceptually doesn't make sense,
so don't allow this.

Signed-off-by: Nicholas Nunley <nicholas.d.nunley@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: Implement DDP support in i40e driver
Aleksandr Loktionov [Wed, 6 Feb 2019 23:08:16 +0000 (15:08 -0800)]
i40e: Implement DDP support in i40e driver

This patch introduces DDP (Dynamic Device Personalization) which allows
loading profiles that change the way internal parser interprets processed
frames. To load DDP profiles it utilizes ethtool flash feature. The files
with recipes must be located in /var/lib/firmware directory. Afterwards
the recipe can be loaded by invoking:

    ethtool -f <if_name> <file_name> 100
    ethtool -f <if_name> - 100

See further details of this feature in the i40e documentation, or
visit
https://www.intel.com/content/www/us/en/architecture-and-technology/ethernet/dynamic-device-personalization-brief.html

The driver shall verify DDP profile can be loaded in accordance with
the rules:
* Package with Group ID 0 are exclusive and can only be loaded the first.
* Packages with Group ID 0x01-0xFE can only be loaded simultaneously
   with the packages from the same group.
* Packages with Group ID 0xFF are compatible with all other packages.

Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoi40e: Queues are reserved despite "Invalid argument" error
Adam Ludkiewicz [Wed, 6 Feb 2019 23:08:15 +0000 (15:08 -0800)]
i40e: Queues are reserved despite "Invalid argument" error

Added a new local variable in the i40e_setup_tc function named
old_queue_pairs so num_queue_pairs can be restored to the correct
value in case configuring queue channels fails. Additionally, moved
the exit label in the i40e_setup_tc function so the if (need_reset)
block can be executed.
Also, fixed data packing in the i40e_setup_tc function.

Signed-off-by: Adam Ludkiewicz <adam.ludkiewicz@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
5 years agoipmi: ipmi_si_hardcode.c: init si_type array to fix a crash
Tony Camuso [Tue, 9 Apr 2019 19:20:03 +0000 (15:20 -0400)]
ipmi: ipmi_si_hardcode.c: init si_type array to fix a crash

The intended behavior of function ipmi_hardcode_init_one() is to default
to kcs interface when no type argument is presented when initializing
ipmi with hard coded addresses.

However, the array of char pointers allocated on the stack by function
ipmi_hardcode_init() was not inited to zeroes, so it contained stack
debris.

Consequently, passing the cruft stored in this array to function
ipmi_hardcode_init_one() caused a crash when it was unable to detect
that the char * being passed was nonsense and tried to access the
address specified by the bogus pointer.

The fix is simply to initialize the si_type array to zeroes, so if
there were no type argument given to at the command line, function
ipmi_hardcode_init_one() could properly default to the kcs interface.

Signed-off-by: Tony Camuso <tcamuso@redhat.com>
Message-Id: <1554837603-40299-1-git-send-email-tcamuso@redhat.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
5 years agoipmi: Fix failure on SMBIOS specified devices
Corey Minyard [Tue, 16 Apr 2019 21:26:45 +0000 (16:26 -0500)]
ipmi: Fix failure on SMBIOS specified devices

An extra memset was put into a place that cleared the interface
type.

Reported-by: Tony Camuso <tcamuso@redhat.com>
Fixes: 3cd83bac481dc4 ("ipmi: Consolidate the adding of platform devices")
Signed-off-by: Corey Minyard <cminyard@mvista.com>
5 years agoMerge tag 'riscv-for-linus-5.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 16 Apr 2019 17:46:37 +0000 (10:46 -0700)]
Merge tag 'riscv-for-linus-5.1-rc6' of git://git./linux/kernel/git/palmer/riscv-linux

Pull RISC-V fixes from Palmer Dabbelt:
 "This contains an assortment of RISC-V-related fixups that we found
  after rc4. They're all really unrelated:

   - The addition of a 32-bit defconfig, to emphasize testing the 32-bit
     port.

   - A device tree bindings patch, which is pre-work for some patches
     that target 5.2.

   - A fix to support booting on systems with more physical memory than
     the maximum supported by the kernel"

* tag 'riscv-for-linus-5.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/palmer/riscv-linux:
  RISC-V: Fix Maximum Physical Memory 2GiB option for 64bit systems
  dt-bindings: clock: sifive: add FU540-C000 PRCI clock constants
  RISC-V: Add separate defconfig for 32bit systems

5 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Tue, 16 Apr 2019 15:52:00 +0000 (08:52 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM fixes from Paolo Bonzini:
 "5.1 keeps its reputation as a big bugfix release for KVM x86.

   - Fix for a memory leak introduced during the merge window

   - Fixes for nested VMX with ept=0

   - Fixes for AMD (APIC virtualization, NMI injection)

   - Fixes for Hyper-V under KVM and KVM under Hyper-V

   - Fixes for 32-bit SMM and tests for SMM virtualization

   - More array_index_nospec peppering"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (21 commits)
  KVM: x86: avoid misreporting level-triggered irqs as edge-triggered in tracing
  KVM: fix spectrev1 gadgets
  KVM: x86: fix warning Using plain integer as NULL pointer
  selftests: kvm: add a selftest for SMM
  selftests: kvm: fix for compilers that do not support -no-pie
  selftests: kvm/evmcs_test: complete I/O before migrating guest state
  KVM: x86: Always use 32-bit SMRAM save state for 32-bit kernels
  KVM: x86: Don't clear EFER during SMM transitions for 32-bit vCPU
  KVM: x86: clear SMM flags before loading state while leaving SMM
  KVM: x86: Open code kvm_set_hflags
  KVM: x86: Load SMRAM in a single shot when leaving SMM
  KVM: nVMX: Expose RDPMC-exiting only when guest supports PMU
  KVM: x86: Raise #GP when guest vCPU do not support PMU
  x86/kvm: move kvm_load/put_guest_xcr0 into atomic context
  KVM: x86: svm: make sure NMI is injected after nmi_singlestep
  svm/avic: Fix invalidate logical APIC id entry
  Revert "svm: Fix AVIC incomplete IPI emulation"
  kvm: mmu: Fix overflow on kvm mmu page limit calculation
  KVM: nVMX: always use early vmcs check when EPT is disabled
  KVM: nVMX: allow tests to use bad virtual-APIC page address
  ...

5 years agoKVM: x86: avoid misreporting level-triggered irqs as edge-triggered in tracing
Vitaly Kuznetsov [Wed, 27 Mar 2019 14:12:20 +0000 (15:12 +0100)]
KVM: x86: avoid misreporting level-triggered irqs as edge-triggered in tracing

In __apic_accept_irq() interface trig_mode is int and actually on some code
paths it is set above u8:

kvm_apic_set_irq() extracts it from 'struct kvm_lapic_irq' where trig_mode
is u16. This is done on purpose as e.g. kvm_set_msi_irq() sets it to
(1 << 15) & e->msi.data

kvm_apic_local_deliver sets it to reg & (1 << 15).

Fix the immediate issue by making 'tm' into u16. We may also want to adjust
__apic_accept_irq() interface and use proper sizes for vector, level,
trig_mode but this is not urgent.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: fix spectrev1 gadgets
Paolo Bonzini [Thu, 11 Apr 2019 09:16:47 +0000 (11:16 +0200)]
KVM: fix spectrev1 gadgets

These were found with smatch, and then generalized when applicable.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: x86: fix warning Using plain integer as NULL pointer
Hariprasad Kelam [Sat, 6 Apr 2019 09:36:58 +0000 (15:06 +0530)]
KVM: x86: fix warning Using plain integer as NULL pointer

Changed passing argument as "0 to NULL" which resolves below sparse warning

arch/x86/kvm/x86.c:3096:61: warning: Using plain integer as NULL pointer

Signed-off-by: Hariprasad Kelam <hariprasad.kelam@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoselftests: kvm: add a selftest for SMM
Vitaly Kuznetsov [Wed, 10 Apr 2019 09:38:33 +0000 (11:38 +0200)]
selftests: kvm: add a selftest for SMM

Add a simple test for SMM, based on VMX.  The test implements its own
sync between the guest and the host as using our ucall library seems to
be too cumbersome: SMI handler is happening in real-address mode.

This patch also fixes KVM_SET_NESTED_STATE to happen after
KVM_SET_VCPU_EVENTS, in fact it places it last.  This is because
KVM needs to know whether the processor is in SMM or not.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoselftests: kvm: fix for compilers that do not support -no-pie
Paolo Bonzini [Thu, 11 Apr 2019 13:51:19 +0000 (15:51 +0200)]
selftests: kvm: fix for compilers that do not support -no-pie

-no-pie was added to GCC at the same time as their configuration option
--enable-default-pie.  Compilers that were built before do not have
-no-pie, but they also do not need it.  Detect the option at build
time.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoselftests: kvm/evmcs_test: complete I/O before migrating guest state
Paolo Bonzini [Thu, 11 Apr 2019 13:57:14 +0000 (15:57 +0200)]
selftests: kvm/evmcs_test: complete I/O before migrating guest state

Starting state migration after an IO exit without first completing IO
may result in test failures.  We already have two tests that need this
(this patch in fact fixes evmcs_test, similar to what was fixed for
state_test in commit 0f73bbc851ed, "KVM: selftests: complete IO before
migrating guest state", 2019-03-13) and a third is coming.  So, move the
code to vcpu_save_state, and while at it do not access register state
until after I/O is complete.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: x86: Always use 32-bit SMRAM save state for 32-bit kernels
Sean Christopherson [Tue, 2 Apr 2019 15:10:48 +0000 (08:10 -0700)]
KVM: x86: Always use 32-bit SMRAM save state for 32-bit kernels

Invoking the 64-bit variation on a 32-bit kenrel will crash the guest,
trigger a WARN, and/or lead to a buffer overrun in the host, e.g.
rsm_load_state_64() writes r8-r15 unconditionally, but enum kvm_reg and
thus x86_emulate_ctxt._regs only define r8-r15 for CONFIG_X86_64.

KVM allows userspace to report long mode support via CPUID, even though
the guest is all but guaranteed to crash if it actually tries to enable
long mode.  But, a pure 32-bit guest that is ignorant of long mode will
happily plod along.

SMM complicates things as 64-bit CPUs use a different SMRAM save state
area.  KVM handles this correctly for 64-bit kernels, e.g. uses the
legacy save state map if userspace has hid long mode from the guest,
but doesn't fare well when userspace reports long mode support on a
32-bit host kernel (32-bit KVM doesn't support 64-bit guests).

Since the alternative is to crash the guest, e.g. by not loading state
or explicitly requesting shutdown, unconditionally use the legacy SMRAM
save state map for 32-bit KVM.  If a guest has managed to get far enough
to handle SMIs when running under a weird/buggy userspace hypervisor,
then don't deliberately crash the guest since there are no downsides
(from KVM's perspective) to allow it to continue running.

Fixes: 660a5d517aaab ("KVM: x86: save/load state on SMM switch")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: x86: Don't clear EFER during SMM transitions for 32-bit vCPU
Sean Christopherson [Tue, 2 Apr 2019 15:10:47 +0000 (08:10 -0700)]
KVM: x86: Don't clear EFER during SMM transitions for 32-bit vCPU

Neither AMD nor Intel CPUs have an EFER field in the legacy SMRAM save
state area, i.e. don't save/restore EFER across SMM transitions.  KVM
somewhat models this, e.g. doesn't clear EFER on entry to SMM if the
guest doesn't support long mode.  But during RSM, KVM unconditionally
clears EFER so that it can get back to pure 32-bit mode in order to
start loading CRs with their actual non-SMM values.

Clear EFER only when it will be written when loading the non-SMM state
so as to preserve bits that can theoretically be set on 32-bit vCPUs,
e.g. KVM always emulates EFER_SCE.

And because CR4.PAE is cleared only to play nice with EFER, wrap that
code in the long mode check as well.  Note, this may result in a
compiler warning about cr4 being consumed uninitialized.  Re-read CR4
even though it's technically unnecessary, as doing so allows for more
readable code and RSM emulation is not a performance critical path.

Fixes: 660a5d517aaab ("KVM: x86: save/load state on SMM switch")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: x86: clear SMM flags before loading state while leaving SMM
Sean Christopherson [Tue, 2 Apr 2019 15:03:11 +0000 (08:03 -0700)]
KVM: x86: clear SMM flags before loading state while leaving SMM

RSM emulation is currently broken on VMX when the interrupted guest has
CR4.VMXE=1.  Stop dancing around the issue of HF_SMM_MASK being set when
loading SMSTATE into architectural state, e.g. by toggling it for
problematic flows, and simply clear HF_SMM_MASK prior to loading
architectural state (from SMRAM save state area).

Reported-by: Jon Doron <arilou@gmail.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: Liran Alon <liran.alon@oracle.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Fixes: 5bea5123cbf0 ("KVM: VMX: check nested state and CR4.VMXE against SMM")
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Tested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: x86: Open code kvm_set_hflags
Sean Christopherson [Tue, 2 Apr 2019 15:03:10 +0000 (08:03 -0700)]
KVM: x86: Open code kvm_set_hflags

Prepare for clearing HF_SMM_MASK prior to loading state from the SMRAM
save state map, i.e. kvm_smm_changed() needs to be called after state
has been loaded and so cannot be done automatically when setting
hflags from RSM.

Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: x86: Load SMRAM in a single shot when leaving SMM
Sean Christopherson [Tue, 2 Apr 2019 15:03:09 +0000 (08:03 -0700)]
KVM: x86: Load SMRAM in a single shot when leaving SMM

RSM emulation is currently broken on VMX when the interrupted guest has
CR4.VMXE=1.  Rather than dance around the issue of HF_SMM_MASK being set
when loading SMSTATE into architectural state, ideally RSM emulation
itself would be reworked to clear HF_SMM_MASK prior to loading non-SMM
architectural state.

Ostensibly, the only motivation for having HF_SMM_MASK set throughout
the loading of state from the SMRAM save state area is so that the
memory accesses from GET_SMSTATE() are tagged with role.smm.  Load
all of the SMRAM save state area from guest memory at the beginning of
RSM emulation, and load state from the buffer instead of reading guest
memory one-by-one.

This paves the way for clearing HF_SMM_MASK prior to loading state,
and also aligns RSM with the enter_smm() behavior, which fills a
buffer and writes SMRAM save state in a single go.

Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: nVMX: Expose RDPMC-exiting only when guest supports PMU
Liran Alon [Mon, 25 Mar 2019 19:09:17 +0000 (21:09 +0200)]
KVM: nVMX: Expose RDPMC-exiting only when guest supports PMU

Issue was discovered when running kvm-unit-tests on KVM running as L1 on
top of Hyper-V.

When vmx_instruction_intercept unit-test attempts to run RDPMC to test
RDPMC-exiting, it is intercepted by L1 KVM which it's EXIT_REASON_RDPMC
handler raise #GP because vCPU exposed by Hyper-V doesn't support PMU.
Instead of unit-test expectation to be reflected with EXIT_REASON_RDPMC.

The reason vmx_instruction_intercept unit-test attempts to run RDPMC
even though Hyper-V doesn't support PMU is because L1 expose to L2
support for RDPMC-exiting. Which is reasonable to assume that is
supported only in case CPU supports PMU to being with.

Above issue can easily be simulated by modifying
vmx_instruction_intercept config in x86/unittests.cfg to run QEMU with
"-cpu host,+vmx,-pmu" and run unit-test.

To handle issue, change KVM to expose RDPMC-exiting only when guest
supports PMU.

Reported-by: Saar Amar <saaramar@microsoft.com>
Reviewed-by: Mihai Carabas <mihai.carabas@oracle.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: x86: Raise #GP when guest vCPU do not support PMU
Liran Alon [Mon, 25 Mar 2019 19:10:17 +0000 (21:10 +0200)]
KVM: x86: Raise #GP when guest vCPU do not support PMU

Before this change, reading a VMware pseduo PMC will succeed even when
PMU is not supported by guest. This can easily be seen by running
kvm-unit-test vmware_backdoors with "-cpu host,-pmu" option.

Reviewed-by: Mihai Carabas <mihai.carabas@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agox86/kvm: move kvm_load/put_guest_xcr0 into atomic context
WANG Chao [Fri, 12 Apr 2019 07:55:39 +0000 (15:55 +0800)]
x86/kvm: move kvm_load/put_guest_xcr0 into atomic context

guest xcr0 could leak into host when MCE happens in guest mode. Because
do_machine_check() could schedule out at a few places.

For example:

kvm_load_guest_xcr0
...
kvm_x86_ops->run(vcpu) {
  vmx_vcpu_run
    vmx_complete_atomic_exit
      kvm_machine_check
        do_machine_check
          do_memory_failure
            memory_failure
              lock_page

In this case, host_xcr0 is 0x2ff, guest vcpu xcr0 is 0xff. After schedule
out, host cpu has guest xcr0 loaded (0xff).

In __switch_to {
     switch_fpu_finish
       copy_kernel_to_fpregs
         XRSTORS

If any bit i in XSTATE_BV[i] == 1 and xcr0[i] == 0, XRSTORS will
generate #GP (In this case, bit 9). Then ex_handler_fprestore kicks in
and tries to reinitialize fpu by restoring init fpu state. Same story as
last #GP, except we get DOUBLE FAULT this time.

Cc: stable@vger.kernel.org
Signed-off-by: WANG Chao <chao.wang@ucloud.cn>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: x86: svm: make sure NMI is injected after nmi_singlestep
Vitaly Kuznetsov [Wed, 3 Apr 2019 14:06:42 +0000 (16:06 +0200)]
KVM: x86: svm: make sure NMI is injected after nmi_singlestep

I noticed that apic test from kvm-unit-tests always hangs on my EPYC 7401P,
the hanging test nmi-after-sti is trying to deliver 30000 NMIs and tracing
shows that we're sometimes able to deliver a few but never all.

When we're trying to inject an NMI we may fail to do so immediately for
various reasons, however, we still need to inject it so enable_nmi_window()
arms nmi_singlestep mode. #DB occurs as expected, but we're not checking
for pending NMIs before entering the guest and unless there's a different
event to process, the NMI will never get delivered.

Make KVM_REQ_EVENT request on the vCPU from db_interception() to make sure
pending NMIs are checked and possibly injected.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agosvm/avic: Fix invalidate logical APIC id entry
Suthikulpanit, Suravee [Tue, 26 Mar 2019 03:57:37 +0000 (03:57 +0000)]
svm/avic: Fix invalidate logical APIC id entry

Only clear the valid bit when invalidate logical APIC id entry.
The current logic clear the valid bit, but also set the rest of
the bits (including reserved bits) to 1.

Fixes: 98d90582be2e ('svm: Fix AVIC DFR and LDR handling')
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoRevert "svm: Fix AVIC incomplete IPI emulation"
Suthikulpanit, Suravee [Wed, 20 Mar 2019 08:12:28 +0000 (08:12 +0000)]
Revert "svm: Fix AVIC incomplete IPI emulation"

This reverts commit bb218fbcfaaa3b115d4cd7a43c0ca164f3a96e57.

As Oren Twaig pointed out the old discussion:

  https://patchwork.kernel.org/patch/8292231/

that the change coud potentially cause an extra IPI to be sent to
the destination vcpu because the AVIC hardware already set the IRR bit
before the incomplete IPI #VMEXIT with id=1 (target vcpu is not running).
Since writting to ICR and ICR2 will also set the IRR. If something triggers
the destination vcpu to get scheduled before the emulation finishes, then
this could result in an additional IPI.

Also, the issue mentioned in the commit bb218fbcfaaa was misdiagnosed.

Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Reported-by: Oren Twaig <oren@scalemp.com>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agokvm: mmu: Fix overflow on kvm mmu page limit calculation
Ben Gardon [Mon, 8 Apr 2019 18:07:30 +0000 (11:07 -0700)]
kvm: mmu: Fix overflow on kvm mmu page limit calculation

KVM bases its memory usage limits on the total number of guest pages
across all memslots. However, those limits, and the calculations to
produce them, use 32 bit unsigned integers. This can result in overflow
if a VM has more guest pages that can be represented by a u32. As a
result of this overflow, KVM can use a low limit on the number of MMU
pages it will allocate. This makes KVM unable to map all of guest memory
at once, prompting spurious faults.

Tested: Ran all kvm-unit-tests on an Intel Haswell machine. This patch
introduced no new failures.

Signed-off-by: Ben Gardon <bgardon@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: nVMX: always use early vmcs check when EPT is disabled
Paolo Bonzini [Mon, 15 Apr 2019 13:57:19 +0000 (15:57 +0200)]
KVM: nVMX: always use early vmcs check when EPT is disabled

The remaining failures of vmx.flat when EPT is disabled are caused by
incorrectly reflecting VMfails to the L1 hypervisor.  What happens is
that nested_vmx_restore_host_state corrupts the guest CR3, reloading it
with the host's shadow CR3 instead, because it blindly loads GUEST_CR3
from the vmcs01.

For simplicity let's just always use hardware VMCS checks when EPT is
disabled.  This way, nested_vmx_restore_host_state is not reached at
all (or at least shouldn't be reached).

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoKVM: nVMX: allow tests to use bad virtual-APIC page address
Paolo Bonzini [Mon, 15 Apr 2019 13:16:17 +0000 (15:16 +0200)]
KVM: nVMX: allow tests to use bad virtual-APIC page address

As mentioned in the comment, there are some special cases where we can simply
clear the TPR shadow bit from the CPU-based execution controls in the vmcs02.
Handle them so that we can remove some XFAILs from vmx.flat.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agobnx2x: fix spelling mistake "dicline" -> "decline"
Colin Ian King [Mon, 15 Apr 2019 15:47:03 +0000 (16:47 +0100)]
bnx2x: fix spelling mistake "dicline" -> "decline"

There is a spelling mistake in a BNX2X_ERR message, fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>