Upon reception of a new frame, the emac driver checks for a number
of error conditions, and flag the packet as "bad" if any of these
are present. It then allocates a skb unconditionally, but only uses
it if the packet is "good". On the error path, the skb is just forgotten,
and the system leaks memory.

The piece of junk I have on my desk seems to encounter such error
frequently enough so that the box goes OOM after a couple of days,
which makes me grumpy.

Fix this by moving the allocation on the "good_packet" path (and
convert it to netdev_alloc_skb while we're at it).

Tested on a random Allwinner A20 board.

Cc: Stefan Roese <sr@denx.de>
Cc: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: <stable@vger.kernel.org> # 3.11+
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Dave reported following splat, caused by improper use of
IP_INC_STATS_BH() in process context.

BUG: using __this_cpu_add() in preemptible [00000000] code: trinity-c117/14551
caller is __this_cpu_preempt_check+0x13/0x20
CPU: 3 PID: 14551 Comm: trinity-c117 Not tainted 3.16.0+ #33
 ffffffff9ec898f0 0000000047ea7e23 ffff88022d32f7f0 ffffffff9e7ee207
 0000000000000003 ffff88022d32f818 ffffffff9e397eaa ffff88023ee70b40
 ffff88022d32f970 ffff8801c026d580 ffff88022d32f828 ffffffff9e397ee3
Call Trace:
 [<ffffffff9e7ee207>] dump_stack+0x4e/0x7a
 [<ffffffff9e397eaa>] check_preemption_disabled+0xfa/0x100
 [<ffffffff9e397ee3>] __this_cpu_preempt_check+0x13/0x20
 [<ffffffffc0839872>] sctp_packet_transmit+0x692/0x710 [sctp]
 [<ffffffffc082a7f2>] sctp_outq_flush+0x2a2/0xc30 [sctp]
 [<ffffffff9e0d985c>] ? mark_held_locks+0x7c/0xb0
 [<ffffffff9e7f8c6d>] ? _raw_spin_unlock_irqrestore+0x5d/0x80
 [<ffffffffc082b99a>] sctp_outq_uncork+0x1a/0x20 [sctp]
 [<ffffffffc081e112>] sctp_cmd_interpreter.isra.23+0x1142/0x13f0 [sctp]
 [<ffffffffc081c86b>] sctp_do_sm+0xdb/0x330 [sctp]
 [<ffffffff9e0b8f1b>] ? preempt_count_sub+0xab/0x100
 [<ffffffffc083b350>] ? sctp_cname+0x70/0x70 [sctp]
 [<ffffffffc08389ca>] sctp_primitive_ASSOCIATE+0x3a/0x50 [sctp]
 [<ffffffffc083358f>] sctp_sendmsg+0x88f/0xe30 [sctp]
 [<ffffffff9e0d673a>] ? lock_release_holdtime.part.28+0x9a/0x160
 [<ffffffff9e0d62ce>] ? put_lock_stats.isra.27+0xe/0x30
 [<ffffffff9e73b624>] inet_sendmsg+0x104/0x220
 [<ffffffff9e73b525>] ? inet_sendmsg+0x5/0x220
 [<ffffffff9e68ac4e>] sock_sendmsg+0x9e/0xe0
 [<ffffffff9e1c0c09>] ? might_fault+0xb9/0xc0
 [<ffffffff9e1c0bae>] ? might_fault+0x5e/0xc0
 [<ffffffff9e68b234>] SYSC_sendto+0x124/0x1c0
 [<ffffffff9e0136b0>] ? syscall_trace_enter+0x250/0x330
 [<ffffffff9e68c3ce>] SyS_sendto+0xe/0x10
 [<ffffffff9e7f9be4>] tracesys+0xdd/0xe2

This is a followup of commits f1d8cba6 ("inet: fix possible
seqlock deadlocks") and 7f88c6b2

 ("ipv6: fix possible seqlock
deadlock in ip6_finish_output2")

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Reported-by: Dave Jones <davej@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit a71e3c37 ("net: phy: Set the driver when registering an MDIO bus
device") caused the following regression on the fec driver:

root@imx6qsabresd:~# echo mem > /sys/power/state
PM: Syncing filesystems ... done.
Freezing user space processes ... (elapsed 0.003 seconds) done.
Freezing remaining freezable tasks ... (elapsed 0.002 seconds) done.
Unable to handle kernel NULL pointer dereference at virtual address 0000002c
pgd = bcd14000
[0000002c] *pgd=4d9e0831, *pte=00000000, *ppte=00000000
Internal error: Oops: 17 [#1] SMP ARM
Modules linked in:
CPU: 0 PID: 617 Comm: sh Not tainted 3.16.0 #17
task: bc0c4e00 ti: bceb6000 task.ti: bceb6000
PC is at fec_suspend+0x10/0x70
LR is at dpm_run_callback.isra.7+0x34/0x6c
pc : [<803f8a98>]    lr : [<80361f44>]    psr: 600f0013
sp : bceb7d70  ip : bceb7d88  fp : bceb7d84
r10: 8091523c  r9 : 00000000  r8 : bd88f478
r7 : 803f8a88  r6 : 81165988  r5 : 00000000  r4 : 00000000
r3 : 00000000  r2 : 00000000  r1 : bd88f478  r0 : bd88f478
Flags: nZCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user
Control: 10c5387d  Table: 4cd1404a  DAC: 00000015
Process sh (pid: 617, stack limit = 0xbceb6240)
Stack: (0xbceb7d70 to 0xbceb8000)
....

The problem with the original commit is explained by Russell King:

"It has the effect (as can be seen from the oops) of attaching the MDIO bus
device (itself is a bus-less device) to the platform driver, which means
that if the platform driver supports power management, it will be called
to power manage the MDIO bus device.

Moreover, drivers do not expect to be called for power management
operations for devices which they haven't probed, and certainly not for
devices which aren't part of the same bus that the driver is registered
against."

This reverts commit a71e3c37

.

Cc: <stable@vger.kernel.org> #3.16
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Rajesh Borundia says:

====================
qlcnic: Bug fixes

The patch series contains following bug fixes.

* Aggregating tx stats in adapter variable was resulting
  in increase of stats when user runs ifconfig command
  and no traffic is running. Instead aggregate tx stats
  in local variable and then assign it to adapter struct
  variable.
* Set_driver_version was called after registering netdev
  which was resulting in a race between FLR in open
  handler and set_driver_version command as open handler
  can be called simulatneously on another cpu even if probe
  is not complete. So call this command before registering
  netdev.
* dcbnl_ops should be initialized before registering netdev
  as they are referenced in open handler.
====================

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

o Initialization of dcbnl_ops after register netdev may result in
  dcbnl_ops not getting set before it is being accessed from open.
  So, moving it before register_netdev.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

o Earlier, set_drv_version was getting called after register_netdev.
  This was resulting in a race between set_drv_version and FLR called
  from open(). Moving set_drv_version before register_netdev avoids
  the race.

o Log response code in error message on CDRP failure.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

o Aggregating tx stats in adapter variable was resulting in
  an increase in stats even after no traffic was run and
  user runs ifconfig/ethtool command.
o qlcnic_update_stats used to accumulate stats in adapter
  struct at each function call, instead accumulate tx stats
  in local variable and then assign it to adapter structure.

Reported-by: Holger Kiehl <holger.kiehl@dwd.de>
Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Antonio Quartulli says:

====================
pull request net: batman-adv 2014-08-04

this is a pull request intended for net.

It just contains a patch by Sven Eckelmann that fixes the
reception of out-of-order fragments. As explained in the
commit message, the issue was due to a wrong assumption
about hlist_for_each_entry() in batadv_frag_insert_packet().
====================

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

batadv_frag_insert_packet was unable to handle out-of-order packets because it
dropped them directly. This is caused by the way the fragmentation lists is
checked for the correct place to insert a fragmentation entry.

The fragmentation code keeps the fragments in lists. The fragmentation entries
are kept in descending order of sequence number. The list is traversed and each
entry is compared with the new fragment. If the current entry has a smaller
sequence number than the new fragment then the new one has to be inserted
before the current entry. This ensures that the list is still in descending
order.

An out-of-order packet with a smaller sequence number than all entries in the
list still has to be added to the end of the list. The used hlist has no
information about the last entry in the list inside hlist_head and thus the
last entry has to be calculated differently. Currently the code assumes that
the iterator variable of hlist_for_each_entry can be used for this purpose
after the hlist_for_each_entry finished. This is obviously wrong because the
iterator variable is always NULL when the list was completely traversed.

Instead the information about the last entry has to be stored in a different
variable.

This problem was introduced in 610bfc6b


("batman-adv: Receive fragmented packets and merge").

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <antonio@meshcoding.com>

Add missing Makefile rule for r8152 driver

In the current kernel the r8152 driver is *never* built because of a missing rule in drivers/net/Makefile, despite being selected as built-in or module in the .config file. There is no error message or warning to indicate that the driver isn't built. This change adds the rule and lets the driver build.

Tested as built-in and module for 3.15.8.

Signed-off by: JF Le Fillatre <jflf-kernel@gmx.com>

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

Pablo Neira Ayuso says:

====================
Netfilter/IPVS fixes for net

The following patchset contains Netfilter/IPVS fixes for your net tree,
they are:

1) Maintain all DSCP and ECN bits for IPv6 tun forwarding. This
   resolves an inconsistency between IPv4 and IPv6 behaviour.
   Patch from Alex Gartrell via Simon Horman.

2) Fix unnoticeable blink in xt_LED when the led-always-blink option is
   used, from Jiri Prchal.

3) Add missing return in nft_del_setelem(), otherwise this results in a
   double call of nft_data_uninit() in the nf_tables code, from Thomas Graf.
====================

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

Fixes: e110861f

 ("net: add a sysctl to reflect the fwmark on replies")
Cc: Lorenzo Colitti <lorenzo@google.com>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This device needs to be reset to recover from a timeout.
Unfortunately this can be handled only at the level of
the subdrivers.

Signed-off-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

Check for cases when the caller requests 0 bytes instead of running off
and dereferencing potentially invalid iovecs.

Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

nft_del_setelem() currently calls nft_data_uninit() twice on the same
key. Once to release the key which is guaranteed to be NFT_DATA_VALUE
and a second time in the error path to which it falls through.

The second call has been harmless so far though because the type
passed is always NFT_DATA_VALUE which is currently a no-op.

Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

When performing segmentation, the mac_len value is copied right
out of the original skb.  However, this value is not always set correctly
(like when the packet is VLAN-tagged) and we'll end up copying a bad
value.

One way to demonstrate this is to configure a VM which tags
packets internally and turn off VLAN acceleration on the forwarding
bridge port.  The packets show up corrupt like this:
16:18:24.985548 52:54:00:ab:be:25 > 52:54:00:26:ce:a3, ethertype 802.1Q
(0x8100), length 1518: vlan 100, p 0, ethertype 0x05e0,
        0x0000:  8cdb 1c7c 8cdb 0064 4006 b59d 0a00 6402 ...|...d@.....d.
        0x0010:  0a00 6401 9e0d b441 0a5e 64ec 0330 14fa ..d....A.^d..0..
        0x0020:  29e3 01c9 f871 0000 0101 080a 000a e833)....q.........3
        0x0030:  000f 8c75 6e65 7470 6572 6600 6e65 7470 ...unetperf.netp
        0x0040:  6572 6600 6e65 7470 6572 6600 6e65 7470 erf.netperf.netp
        0x0050:  6572 6600 6e65 7470 6572 6600 6e65 7470 erf.netperf.netp
        0x0060:  6572 6600 6e65 7470 6572 6600 6e65 7470 erf.netperf.netp
        ...

This also leads to awful throughput as GSO packets are dropped and
cause retransmissions.

The solution is to set the mac_len using the values already available
in then new skb.  We've already adjusted all of the header offset, so we
might as well correctly figure out the mac_len using skb_reset_mac_len().
After this change, packets are segmented correctly and performance
is restored.

CC: Eric Dumazet <edumazet@google.com>
Signed-off-by: Vlad Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

David Vrabel says:

====================
xen-netfront: more multiqueue fixes

A few more xen-netfront fixes for the multiqueue support added in
3.16-rc1.  It would be great if these could make it into 3.16 but I
suspect it's a little late for that now.

The second patch fixes a significant resource leak that prevents
guests from migrating more than a handful of times.

These have been tested by repeatedly migrating a guest over 250 times
(it would previously fail with this guest after only 8 iterations).
====================

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

When less than the requested number of queues could be created, include
the actual number in the warning (instead of the requested number).

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Since netfront may reconnect to a backend with a different number of
queues, all per-queue Rx and Tx resources (skbs and grant references)
should be freed when disconnecting.

Without this fix, the Tx and Rx grant refs are not released and
netfront will exhaust them after only a few reconnections.  netfront
will fail to connect when no free grant references are available.

Since all Rx bufs are freed and reallocated instead of reused this
will add some additional delay to the reconnection but this is
expected to be small compared to the time taken by any backend hotplug
scripts etc.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

If no queues could be created when connecting to the backend, one of the
error paths would deadlock.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Macvlan devices do not initialize vlan_features.  As a result,
any vlan devices configured on top of macvlans perform very poorly.
Initialize vlan_features based on the vlan features of the lower-level
device.

Signed-off-by: Vlad Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

An FDB entry with vlan_id 0 doesn't mean it is used in vlan 0, but used when
vlan_filtering is disabled.

There is inconsistency around NDA_VLAN whose payload is 0 - even if we add
an entry by RTM_NEWNEIGH without any NDA_VLAN, and even though adding an
entry with NDA_VLAN 0 is prohibited, we get an entry with NDA_VLAN 0 by
RTM_GETNEIGH.

Dumping an FDB entry with vlan_id 0 shouldn't include NDA_VLAN.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>

Otherwise the name of the kobject isn't getting freed and other stuff from
kobject_cleanup() isn't getting called. kobject_put() will call
kobject_del() on its own in kobject_cleanup().

CC: Jay Vosburgh <j.vosburgh@gmail.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The recent commit "e29aa339

 bna: Enable Multi Buffer RX" is causing
a performance regression. It does not properly update 'cmpl' pointer
at the end of the loop in NAPI handler bnad_cq_process(). The result is
only one packet / per NAPI-schedule is processed.

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In vegas we do a multiplication of the cwnd and the rtt. This
may overflow and thus their result is stored in a u64. However, we first
need to cast the cwnd so that actually 64-bit arithmetic is done.

Then, we need to do do_div to allow this to be used on 32-bit arches.

Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: David Laight <David.Laight@ACULAB.COM>
Cc: Doug Leith <doug.leith@nuim.ie>
Fixes: 8d3a564d

 (tcp: tcp_vegas cong avoid fix)
Signed-off-by: Christoph Paasch <christoph.paasch@uclouvain.be>
Signed-off-by: David S. Miller <davem@davemloft.net>

In veno we do a multiplication of the cwnd and the rtt. This
may overflow and thus their result is stored in a u64. However, we first
need to cast the cwnd so that actually 64-bit arithmetic is done.

A first attempt at fixing 76f10177 ([TCP]: TCP Veno congestion
control) was made by 15913114 (tcp: Overflow bug in Vegas), but it
failed to add the required cast in tcp_veno_cong_avoid().

Fixes: 76f10177

 ([TCP]: TCP Veno congestion control)
Signed-off-by: Christoph Paasch <christoph.paasch@uclouvain.be>
Signed-off-by: David S. Miller <davem@davemloft.net>

Ipv4 tunnels created with "local any remote $ip" didn't work properly since
7d442fab

 (ipv4: Cache dst in tunnels). 99% of packets sent via those tunnels
had src addr = 0.0.0.0. That was because only dst_entry was cached, although
fl4.saddr has to be cached too. Every time ip_tunnel_xmit used cached dst_entry
(tunnel_rtable_get returned non-NULL), fl4.saddr was initialized with
tnl_params->saddr (= 0 in our case), and wasn't changed until iptunnel_xmit().

This patch adds saddr to ip_tunnel->dst_cache, fixing this issue.

Reported-by: Sergey Popov <pinkbyte@gentoo.org>
Signed-off-by: Dmitry Popov <ixaphire@qrator.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

A driver should fill magic field of ethtool_eeprom struct in .get_eeprom
and validate it in .set_eeprom. The bna incorrectly validates it in both
and this makes its .get_eeprom interface unusable.

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull Exynos platform DT fix from Grant Likely:
 "Device tree Exynos bug fix for v3.16-rc7

  This bug fix has been brewing for a while.  I hate sending it to you
  so late, but I only got confirmation that it solves the problem this
  past weekend.  The diff looks big for a bug fix, but the majority of
  it is only executed in the Exynos quirk case.  Unfortunately it
  required splitting early_init_dt_scan() in two and adding quirk
  handling in the middle of it on ARM.

  Exynos has buggy firmware that puts bad data into the memory node.
  Commit 1c2f87c2 ("ARM: Get rid of meminfo") exposed the bug by
  dropping the artificial upper bound on the number of memory banks that
  can be added.  Exynos fails to boot after that commit.  This branch
  fixes it by splitting the early DT parse function and inserting a
  fixup hook.  Exynos uses the hook to correct the DT before parsing
  memory regions"

* tag 'devicetree-for-linus' of git://git.secretlab.ca/git/linux:
  arm: Add devicetree fixup machine function
  of: Add memory limiting function for flattened devicetrees
  of: Split early_init_dt_scan into two parts

Pull Xen fix from David Vrabel:
 "Fix BUG when trying to expand the grant table.  This seems to occur
  often during boot with Ubuntu 14.04 PV guests"

* tag 'stable/for-linus-3.16-rc7-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
  x86/xen: safely map and unmap grant frames when in atomic context

Pull KVM fix from Paolo Bonzini:
 "Fix a bug which allows KVM guests to bring down the entire system on
  some 64K enabled ARM64 hosts"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  kvm: arm64: vgic: fix hyp panic with 64k pages on juno platform

This reverts commit 20fbe3ae

.

As reported by Stephen Rothwell, it causes compile failures in certain
configurations:

  drivers/net/usb/cdc_subset.c:360:15: error: 'dummy_prereset' undeclared here (not in a function)
    .pre_reset = dummy_prereset,
                 ^
  drivers/net/usb/cdc_subset.c:361:16: error: 'dummy_postreset' undeclared here (not in a function)
    .post_reset = dummy_postreset,
                  ^

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Acked-by: David Miller <davem@davemloft.net>
Cc: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull networking fixes from David Miller:

 1) Make fragmentation IDs less predictable, from Eric Dumazet.

 2) TSO tunneling can crash in bnx2x driver, fix from Dmitry Kravkov.

 3) Don't allow NULL msg->msg_name just because msg->msg_namelen is
    non-zero, from Andrey Ryabinin.

 4) ndm->ndm_type set using wrong macros, from Jun Zhao.

 5) cdc-ether devices can come up with entries in their address filter,
    so explicitly clear the filter after the device initializes.  From
    Oliver Neukum.

 6) Forgotten refcount bump in xfrm_lookup(), from Steffen Klassert.

 7) Short packets not padded properly, exposing random data, in bcmgenet
    driver.  Fix from Florian Fainelli.

 8) xgbe_probe() doesn't return an error code, but rather zero, when
    netif_set_real_num_tx_queues() fails.  Fix from Wei Yongjun.

 9) USB speed not probed properly in r8152 driver, from Hayes Wang.

10) Transmit logic choosing the outgoing port in the sunvnet driver
    needs to consider a) is the port actually up and b) whether it is a
    switch port.  Fix from David L Stevens.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (27 commits)
  net: phy: re-apply PHY fixups during phy_register_device
  cdc-ether: clean packet filter upon probe
  cdc_subset: deal with a device that needs reset for timeout
  net: sendmsg: fix NULL pointer dereference
  isdn/bas_gigaset: fix a leak on failure path in gigaset_probe()
  ip: make IP identifiers less predictable
  neighbour : fix ndm_type type error issue
  sunvnet: only use connected ports when sending
  can: c_can_platform: Fix raminit, use devm_ioremap() instead of devm_ioremap_resource()
  bnx2x: fix crash during TSO tunneling
  r8152: fix the checking of the usb speed
  net: phy: Ensure the MDIO bus module is held
  net: phy: Set the driver when registering an MDIO bus device
  bnx2x: fix set_setting for some PHYs
  hyperv: Fix error return code in netvsc_init_buf()
  amd-xgbe: Fix error return code in xgbe_probe()
  ath9k: fix aggregation session lockup
  net: bcmgenet: correctly pad short packets
  net: sctp: inherit auth_capable on INIT collisions
  mac80211: fix crash on getting sta info with uninitialized rate control
  ...

arch_gnttab_map_frames() and arch_gnttab_unmap_frames() are called in
atomic context but were calling alloc_vm_area() which might sleep.

Also, if a driver attempts to allocate a grant ref from an interrupt
and the table needs expanding, then the CPU may already by in lazy MMU
mode and apply_to_page_range() will BUG when it tries to re-enable
lazy MMU mode.

These two functions are only used in PV guests.

Introduce arch_gnttab_init() to allocates the virtual address space in
advance.

Avoid the use of apply_to_page_range() by using saving and using the
array of PTE addresses from the alloc_vm_area() call (which ensures
that the required page tables are pre-allocated).

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

If the physical address of GICV isn't page-aligned, then we end up
creating a stage-2 mapping of the page containing it, which causes us to
map neighbouring memory locations directly into the guest.

As an example, consider a platform with GICV at physical 0x2c02f000
running a 64k-page host kernel. If qemu maps this into the guest at
0x80010000, then guest physical addresses 0x80010000 - 0x8001efff will
map host physical region 0x2c020000 - 0x2c02efff. Accesses to these
physical regions may cause UNPREDICTABLE behaviour, for example, on the
Juno platform this will cause an SError exception to EL3, which brings
down the entire physical CPU resulting in RCU stalls / HYP panics / host
crashing / wasted weeks of debugging.

SBSA recommends that systems alias the 4k GICV across the bounding 64k
region, in which case GICV physical could be described as 0x2c020000 in
the above scenario.

This patch fixes the problem by failing the vgic probe if the physical
base address or the size of GICV aren't page-aligned. Note that this
generated a warning in dmesg about freeing enabled IRQs, so I had to
move the IRQ enabling later in the probe.

Cc: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Joel Schopp <joel.schopp@amd.com>
Cc: Don Dutile <ddutile@redhat.com>
Acked-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Joel Schopp <joel.schopp@amd.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

Commit 1c2f87c2


(ARM: 8025/1: Get rid of meminfo) dropped the upper bound on
the number of memory banks that can be added as there was no
technical need in the kernel. It turns out though, some bootloaders
(specifically the arndale-octa exynos boards) may pass invalid memory
information and rely on the kernel to not parse this data. This is a
bug in the bootloader but we still need to work around this.
Work around this by introducing a dt_fixup function. This function
gets called before the flattened devicetree is scanned for memory
and the like. In this fixup function for exynos, limit the maximum
number of memory regions in the devicetree.

Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Tested-by: Andreas Färber <afaerber@suse.de>
[glikely: Added a comment and fixed up function name]
Signed-off-by: Grant Likely <grant.likely@linaro.org>

Buggy bootloaders may pass bogus memory entries in the devicetree.
Add of_fdt_limit_memory to add an upper bound on the number of
entries that can be present in the devicetree.

Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Tested-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Grant Likely <grant.likely@linaro.org>

Currently, early_init_dt_scan validates the header, sets the
boot params, and scans for chosen/memory all in one function.
Split this up into two separate functions (validation/setting
boot params in one, scanning in another) to allow for
additional setup between boot params and scanning the memory.

Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Tested-by: Andreas Färber <afaerber@suse.de>
[glikely: s/early_init_dt_scan_all/early_init_dt_scan_nodes/]
Signed-off-by: Grant Likely <grant.likely@linaro.org>

Commit 87aa9f9c

 ("net: phy: consolidate PHY reset in phy_init_hw()")
moved the call to phy_scan_fixups() in phy_init_hw() after a software
reset is performed.

By the time phy_init_hw() is called in phy_device_register(), no driver
has been bound to this PHY yet, so all the checks in phy_init_hw()
against the PHY driver and the PHY driver's config_init function will
return 0. We will therefore never call phy_scan_fixups() as we should.

Fix this by calling phy_scan_fixups() and check for its return value to
restore the intended functionality.

This broke PHY drivers which do register an early PHY fixup callback to
intercept the PHY probing and do things like changing the 32-bits unique
PHY identifier when a pseudo-PHY address has been used, as well as
board-specific PHY fixups that need to be applied during driver probe
time.

Reported-by: Hauke Merthens <hauke-m@hauke-m.de>
Reported-by: Jonas Gorski <jogo@openwrt.org>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There are devices that don't do reset all the way. So the packet filter should
be set to a sane initial value. Failure to do so leads to intermittent failures
of DHCP on some systems under some conditions.

Signed-off-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>