outer_header_zero() routine checks if the outer_headers match of a
flow-table entry are all zero.

This function uses the size of whole fte_match_param, instead of just
the outer_headers member, causing failure to detect all-zeros if
any other members of the fte_match_param are non-zero.

Use the correct size for zero check.

Fixes: 6dc6071c ("net/mlx5e: Add ethtool flow steering support")
Signed-off-by: Ilan Tayari <ilant@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

On interface remove, the clean-up was done incorrectly causing
an error in the log:
"SET_FLOW_TABLE_ROOT(0x92f) op_mod(0x0) failed...syndrome (0x7e9f14)"

This was caused by the following flow:
-ndo_uninit:
 Move QP state to RST (this disconnects the QP from FT),
 the QP cannot be attached to any FT unless it is in RTS.

-mlx5_rdma_netdev_free:
 cleanup_rx: Destroy FT
 cleanup_tx: Destroy QP and remove QPN from FT

This caused a problem when destroying current FT we tried to
re-attach the QP to the next FT which is not needed.

The correct flow is:
-mlx5_rdma_netdev_free:
	cleanup_rx: remove QPN from FT & Destroy FT
	cleanup_tx: Destroy QP

Fixes: 50854114 ("net/mlx5: Use underlay QPN from the root name space")
Signed-off-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

When driver fail to allocate an entry to send command to FW, it must
notify the calling function and release the memory allocated for
this command.

Fixes: e126ba97 ('mlx5: Add driver for Mellanox Connect-IB adapters')
Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Cc: kernel-team@fb.com
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

Completion on timeout should not free the driver command entry structure
as it will need to access it again once real completion event from FW
will occur.

Fixes: 73dd3a48 ('net/mlx5: Avoid using pending command interface slots')
Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Cc: kernel-team@fb.com
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

The tx_enabled lag event field is used to determine whether a slave is
active.
Current logic uses this value only if the mode is active-backup.

However, LACP mode, although considered a load balancing mode, can mark
a slave as inactive in certain situations (e.g., LACP timeout).

This fix takes the tx_enabled value into account when remapping, with
no respect to the LAG mode (this should not affect the behavior in XOR
mode, since in this mode both slaves are marked as active).

Fixes: 7907f23a (net/mlx5: Implement RoCE LAG feature)
Signed-off-by: Aviv Heller <avivh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

Upon sriov enable, eswitch is always enabled.
Currently, if enable hca failed over all VFs, we would skip eswitch
disable as part of sriov disable, which will lead to resources leak.

Fix it by disabling eswitch if it was enabled (use indication from
eswitch mode).

Fixes: 6b6adee3 ('net/mlx5: SRIOV core code refactoring')
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Noa Osherovich <noaos@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

Commit b1f5bfc2 ("sctp: don't dereference ptr before leaving
_sctp_walk_{params, errors}()") tried to fix the issue that it
may overstep the chunk end for _sctp_walk_{params, errors} with
'chunk_end > offset(length) + sizeof(length)'.

But it introduced a side effect: When processing INIT, it verifies
the chunks with 'param.v == chunk_end' after iterating all params
by sctp_walk_params(). With the check 'chunk_end > offset(length)
+ sizeof(length)', it would return when the last param is not yet
accessed. Because the last param usually is fwdtsn supported param
whose size is 4 and 'chunk_end == offset(length) + sizeof(length)'

This is a badly issue even causing sctp couldn't process 4-shakes.
Client would always get abort when connecting to server, due to
the failure of INIT chunk verification on server.

The patch is to use 'chunk_end <= offset(length) + sizeof(length)'
instead of 'chunk_end < offset(length) + sizeof(length)' for both
_sctp_walk_params and _sctp_walk_errors.

Fixes: b1f5bfc2 ("sctp: don't dereference ptr before leaving _sctp_walk_{params, errors}()")
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In dccp_feat_init, when ccid_get_builtin_ccids failsto alloc
memory for rx.val, it should free tx.val before returning an
error.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The patch "dccp: fix a memleak that dccp_ipv6 doesn't put reqsk
properly" fixed reqsk refcnt leak for dccp_ipv6. The same issue
exists on dccp_ipv4.

This patch is to fix it for dccp_ipv4.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In dccp_v6_conn_request, after reqsk gets alloced and hashed into
ehash table, reqsk's refcnt is set 3. one is for req->rsk_timer,
one is for hlist, and the other one is for current using.

The problem is when dccp_v6_conn_request returns and finishes using
reqsk, it doesn't put reqsk. This will cause reqsk refcnt leaks and
reqsk obj never gets freed.

Jianlin found this issue when running dccp_memleak.c in a loop, the
system memory would run out.

dccp_memleak.c:
  int s1 = socket(PF_INET6, 6, IPPROTO_IP);
  bind(s1, &sa1, 0x20);
  listen(s1, 0x9);
  int s2 = socket(PF_INET6, 6, IPPROTO_IP);
  connect(s2, &sa1, 0x20);
  close(s1);
  close(s2);

This patch is to put the reqsk before dccp_v6_conn_request returns,
just as what tcp_conn_request does.
Reported-by: Jianlin Shi <jishi@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The buffer passed to bpf_obj_get_info_by_fd() should be initialized
to zeros.  Kernel will enforce that to guarantee we can safely extend
info structures in the future.

Making the bpf_obj_get_info_by_fd() call in libbpf perform the zeroing
is problematic, however, since some members of the info structures
may need to be initialized by the callers (for instance pointers
to buffers to which kernel is to dump translated and jited images).

Remove the zeroing and fix up the in-tree callers before any kernel
has been released with this code.

As Daniel points out this seems to be the intended operation anyway,
since commit 95b9afd3 ("bpf: Test for bpf ID") is itself setting
the buffer pointers before calling bpf_obj_get_info_by_fd().
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Apparently netpoll_setup() assumes that netpoll.dev_name is a pointer
when checking if the device name is set:

if (np->dev_name) {
  ...

However the field is a character array, therefore the condition always
yields true. Check instead whether the first byte of the array has a
non-zero value.
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit de77ecd4 ("bonding: improve link-status update in mii-monitoring")
moves link status commitment into bond_mii_monitor(), but it still relies
on the return value of bond_miimon_inspect() as the hint. We need to return
non-zero as long as we propose a link status change.

Fixes: de77ecd4 ("bonding: improve link-status update in mii-monitoring")
Reported-by: Benjamin Gilbert <benjamin.gilbert@coreos.com>
Tested-by: Benjamin Gilbert <benjamin.gilbert@coreos.com>
Cc: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We must use pre-processor conditional block or suitable accessors to
manipulate skb->sp elsewhere builds lacking the CONFIG_XFRM will break.

Fixes: dce4551c ("udp: preserve head state for IP_CMSG_PASSSEC")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Before commit bf8f6952 ("Add blurb about RGMII") it was unclear
whose responsibility it was to insert the required clock skew, and
in hindsight, some PHY drivers got it wrong. The solution forward
is to introduce a new property, explicitly requiring skew from the
node to which it is attached. In the interim, this driver will handle
all 4 RGMII modes identically (no skew).

Fixes: 52dfc830 ("net: ethernet: add driver for Aurora VLSI NB8800 Ethernet controller")
Signed-off-by: Marc Gonzalez <marc_gonzalez@sigmadesigns.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The logic for computing page buffer scatter does not take into
account the impact of compound pages. Therefore the optimization
to compute number of slots was incorrect and could cause stack
corruption a skb was sent with lots of fragments from huge pages.

This reverts commit 60b86665.

Fixes: 60b86665 ("netvsc: optimize calculation of number of slots")
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The error paths set err, but it's not returned.

I wondered if we should fix all of the callers to check the returned
value, but Ben explains why the code is this way:

> Most call sites ignore it on purpose. There's nothing we can do if
> we fail to get a buffer at interrupt time, so we point the buffer to
> the scratch page so the HW doesn't DMA into lalaland and lose the
> packet.
>
> The one call site that tests and can fail is the one used when brining
> the interface up. If we fail to allocate at that point, we fail the
> ifup. But as you noticed, I do have a bug not returning the error.
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

RFC 2465 defines ipv6IfStatsOutFragFails as:

	"The number of IPv6 datagrams that have been discarded
	 because they needed to be fragmented at this output
	 interface but could not be."

The existing implementation, instead, would increase the counter
twice in case we fail to allocate room for single fragments:
once for the fragment, once for the datagram.

This didn't look intentional though. In one of the two affected
affected failure paths, the double increase was simply a result
of a new 'goto fail' statement, introduced to avoid a skb leak.
The other path appears to be affected since at least 2.6.12-rc2.
Reported-by: Sabrina Dubroca <sdubroca@redhat.com>
Fixes: 1d325d21 ("ipv6: ip6_fragment: fix headroom tests and skb leak")
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

I forgot one spot when introducing struct test_obj_val.

Fixes: e859afe1 ("lib: test_rhashtable: fix for large entry counts")
Reported by: kernel test robot <fengguang.wu@intel.com>
Signed-off-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit e5a03bfd ("phy: Add an mdio_device structure")
introduced a spurious trailing semicolon. Remove it.
Signed-off-by: Marc Gonzalez <marc_gonzalez@sigmadesigns.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Paul Moore reported a SELinux/IP_PASSSEC regression
caused by missing skb->sp at recvmsg() time. We need to
preserve the skb head state to process the IP_CMSG_PASSSEC
cmsg.

With this commit we avoid releasing the skb head state in the
BH even if a secpath is attached to the current skb, and stores
the skb status (with/without head states) in the scratch area,
so that we can access it at skb deallocation time, without
incurring in cache-miss penalties.

This also avoids misusing the skb CB for ipv6 packets,
as introduced by the commit 0ddf3fb2 ("udp: preserve
skb->dst if required for IP options processing").

Clean a bit the scratch area helpers implementation, to
reduce the code differences between 32 and 64 bits build.
Reported-by: Paul Moore <paul@paul-moore.com>
Fixes: 0a463c78 ("udp: avoid a cache miss on dequeue")
Fixes: 0ddf3fb2 ("udp: preserve skb->dst if required for IP options processing")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Tested-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The mt7530 driver has its dsa_switch_ops::get_tag_protocol function
check ds->cpu_port_mask to issue a warning in case the configured CPU
port is not capable of supporting tags.

After commit 14be36c2 ("net: dsa: Initialize all CPU and enabled
ports masks in dsa_ds_parse()") we slightly re-arranged the
initialization such that this was no longer working. Just make sure that
ds->cpu_port_mask is set prior to the first call to get_tag_protocol,
thus restoring the expected contract. In case of error, the CPU port bit
is cleared.

Fixes: 14be36c2 ("net: dsa: Initialize all CPU and enabled ports masks in dsa_ds_parse()")
Reported-by: Sean Wang <sean.wang@mediatek.com>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There are multiple reports showing we have a use-after-free in
the timer prb_retire_rx_blk_timer_expired(), where we use struct
tpacket_kbdq_core::pkbdq, a pg_vec, after it gets freed by
free_pg_vec().

The interesting part is it is not freed via packet_release() but
via packet_setsockopt(), which means we are not closing the socket.
Looking into the big and fat function packet_set_ring(), this could
happen if we satisfy the following conditions:

1. closing == 0, not on packet_release() path
2. req->tp_block_nr == 0, we don't allocate a new pg_vec
3. rx_ring->pg_vec is already set as V3, which means we already called
   packet_set_ring() wtih req->tp_block_nr > 0 previously
4. req->tp_frame_nr == 0, pass sanity check
5. po->mapped == 0, never called mmap()

In this scenario we are clearing the old rx_ring->pg_vec, so we need
to free this pg_vec, but we don't stop the timer on this path because
of closing==0.

The timer has to be stopped as long as we need to free pg_vec, therefore
the check on closing!=0 is wrong, we should check pg_vec!=NULL instead.

Thanks to liujian for testing different fixes.

Reported-by: alexander.levin@verizon.com
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Reported-by: liujian (CE) <liujian56@huawei.com>
Tested-by: liujian (CE) <liujian56@huawei.com>
Cc: Ding Tianhong <dingtianhong@huawei.com>
Cc: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Populate mii_bus->parent with our own platform device before
registering, which makes it easier to locate the MDIO bus
in sysfs when trying to diagnose problems.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

We had reports of 50us not being sufficient to reset the MAC,
thus hitting the "Hardware reset failed" error bringing the
interface up on some AST2400 based machines.

This bumps the timeout to 200us.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Before the 'type' is validated, we shouldn't use it to fetch the
ovs_ct_attr_lens's minlen and maxlen, else, out of bound access
may happen.

Fixes: 7f8a436e ("openvswitch: Add conntrack action")
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

DMA transfers are not allowed to buffers that are on the stack.
Therefore allocate a buffer to store the result of usb_control_message().

Fixes these bugreports:
https://bugzilla.kernel.org/show_bug.cgi?id=195217

https://bugzilla.redhat.com/show_bug.cgi?id=1421387
https://bugzilla.redhat.com/show_bug.cgi?id=1427398

Shortened kernel backtrace from 4.11.9-200.fc25.x86_64:
kernel: ------------[ cut here ]------------
kernel: WARNING: CPU: 3 PID: 2957 at drivers/usb/core/hcd.c:1587
kernel: transfer buffer not dma capable
kernel: Call Trace:
kernel: dump_stack+0x63/0x86
kernel: __warn+0xcb/0xf0
kernel: warn_slowpath_fmt+0x5a/0x80
kernel: usb_hcd_map_urb_for_dma+0x37f/0x570
kernel: ? try_to_del_timer_sync+0x53/0x80
kernel: usb_hcd_submit_urb+0x34e/0xb90
kernel: ? schedule_timeout+0x17e/0x300
kernel: ? del_timer_sync+0x50/0x50
kernel: ? __slab_free+0xa9/0x300
kernel: usb_submit_urb+0x2f4/0x560
kernel: ? urb_destroy+0x24/0x30
kernel: usb_start_wait_urb+0x6e/0x170
kernel: usb_control_msg+0xdc/0x120
kernel: mcs_get_reg+0x36/0x40 [mcs7780]
kernel: mcs_net_open+0xb5/0x5c0 [mcs7780]
...

Regression goes back to 4.9, so it's a good candidate for -stable.
Though it's the decision of the maintainer.

Thanks to Dan Williams for adding the "transfer buffer not dma capable"
warning in the first place. It instantly pointed me in the right direction.

Patch has been tested with transferring data from a Polar watch.
Signed-off-by: Thomas Jarosch <thomas.jarosch@intra2net.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The commit fbf68229 ("net: stmmac: unify registers dumps methods")

in the Linux kernel modified the register dump to store the DMA registers
at the DMA register offset (0x1000) but ethtool (stmmac.c) looks for the
DMA registers after the MAC registers which is offset 55.
This patch copies the DMA registers from the higher offset to the offset
where ethtool expects them.
Signed-off-by: Thor Thayer <thor.thayer@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The commit ffb07550 ("copy_msghdr_from_user(): get rid of
field-by-field copyin") introduce a new sparse warning:

net/socket.c:1919:27: warning: incorrect type in assignment (different address spaces)
net/socket.c:1919:27:    expected void *msg_control
net/socket.c:1919:27:    got void [noderef] <asn:1>*[addressable] msg_control

and a line above 80 chars, let's fix them

Fixes: ffb07550 ("copy_msghdr_from_user(): get rid of field-by-field copyin")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'wireless-drivers-for-davem-2017-07-21' of git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/wireless-drivers

Kalle Valo says:

====================
wireless-drivers fixes for 4.13

Important, but small in size, fixes.

brcmfmac

* fix a regression in SDIO support introduced in v4.13-rc1

rtlwifi

* fix a regression in bluetooth coexistance introduced in v4.13-rc1

iwlwifi

* a few NULL pointer dereferences in the recovery flow

* a small but important fix for IBSS

* a one-liner fix for tracing, which was including too much data
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Each FIB node holds a linked list of routes sharing the same prefix and
length. In the case of IPv4 it's ordered according to table ID, metric
and TOS and only the first route in the list is actually programmed to
the device.

In case a gatewayed route is added somewhere in the list, then after its
nexthop group will be refreshed and become valid (due to the resolution
of its gateway), it'll mistakenly overwrite the existing entry.

Example:
192.168.200.0/24 dev enp3s0np3 scope link metric 1000 offload
192.168.200.0/24 via 192.168.100.1 dev enp3s0np3 metric 1000 offload

Both routes are marked as offloaded despite the fact only the first one
should actually be present in the device's table.

When refreshing the nexthop group, don't write the route to the device's
table unless it's the first in its node.

Fixes: 9aecce1c ("mlxsw: spectrum_router: Correctly handle identical routes")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

During concurrent access testing, threadfunc() concatenated thread ID
and object index to create a unique key like so:

| tdata->objs[i].value = (tdata->id << 16) | i;

This breaks if a user passes an entries parameter of 64k or higher,
since 'i' might use more than 16 bits then. Effectively, this will lead
to duplicate keys in the table.

Fix the problem by introducing a struct holding object and thread ID and
using that as key instead of a single integer type field.

Fixes: f4a3e90b ("rhashtable-test: extend to test concurrency")
Reported by: Manuel Messner <mm@skelett.io>
Signed-off-by: Phil Sutter <phil@nwl.cc>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

Edward Cree says:

====================
bpf: fix verifier min/max handling in BPF_SUB

I managed to come up with a test for the swapped bounds in BPF_SUB, so here
 it is along with a patch that fixes it, separated out from my 'rewrite
 everything' series so it can go to -stable.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

We have to subtract the src max from the dst min, and vice-versa, since
 (e.g.) the smallest result comes from the largest subtrahend.

Fixes: 48461135 ("bpf: allow access into map value arrays")
Signed-off-by: Edward Cree <ecree@solarflare.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

There is a bug in the verifier's handling of BPF_SUB: [a,b] - [c,d] yields
 was [a-c, b-d] rather than the correct [a-d, b-c].  So here is a test
 which, with the bogus handling, will produce ranges of [0,0] and thus
 allowed accesses; whereas the correct handling will give a range of
 [-255, 255] (and hence the right-shift will give a range of [0, 255]) and
 the accesses will be rejected.
Signed-off-by: Edward Cree <ecree@solarflare.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

The mediatek ethernet driver uses interrupts but does not explicitly
include linux/interrupt.h, relying on implicit includes.  Fix this so we
don't get build breaks as happened for ARM in next-20170720.
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

There is some codes of tun/tap module which did not check the return
value of register_netdevice_notifier. Add the check now.
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Potential dangerous invalid memory might be accessed if invalid mac value
reflected from the forward port field in rxd4 caused by possible potential
hardware defects. So added a simple sanity checker to avoid the kind of
situation happening.
Signed-off-by: Sean Wang <sean.wang@mediatek.com>
Acked-by: John Crispin <john@phrozen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'iwlwifi-for-kalle-2017-07-21' of git://git.kernel.org/pub/scm/linux/kernel/git/iwlwifi/iwlwifi-fixes

Some iwlwifi fixes for 4.13:

* A few NULL pointer dereferences in the recovery flow;
* A small but important fix for IBSS;
* A one-liner fix for tracing, which was including too much data;
* Some of these are bugzilla bug fixes;

A hardware/firmware error may happen at any point in time. In
particular, it might happen while mac80211 is in the middle of
a flow. We observed the following situation:
 * mac80211 is in authentication flow, in ieee80211_prep_connection()
 * iwlwifi firmware crashes, but no error can be reported at this
   precise point (mostly because the driver method is void, but even
   if it wasn't we'd just shift to a race condition)
 * mac80211 continues the flow, trying to add the AP station
 * iwlwifi has already set its internal restart flag, and so thinks
   that adding the station is part of the restart and already set up,
   so it uses the information that's supposed to already be in the
   struct

This can happen with any flow in mac80211 and with any information
we try to preserve across hardware restarts.

To fix this, only set a new HW_RESTART_REQUESTED flag and translate
that to IN_HW_RESTART once mac80211 actually starts the restart by
calling our start() method. As a consequence, any mac80211 flow in
progress at the time of the restart will properly finish (certainly
with errors), before the restart is attempted.

This fixes https://bugzilla.kernel.org/show_bug.cgi?id=195299.
Reported-by: djagoo <dev@djagoo.io>
Reported-by: Łukasz Siudut <lsiudut@gmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>