Commit 3f8fd02b1bf1 ("mm/vmalloc: Sync unmappings in
__purge_vmap_area_lazy()") introduced a call to vmalloc_sync_all() in
the vunmap() code-path.  While this change was necessary to maintain
correctness on x86-32-pae kernels, it also adds additional cycles for
architectures that don't need it.

Specifically on x86-64 with CONFIG_VMAP_STACK=y some people reported
severe performance regressions in micro-benchmarks because it now also
calls the x86-64 implementation of vmalloc_sync_all() on vunmap().  But
the vmalloc_sync_all() implementation on x86-64 is only needed for newly
created mappings.

To avoid the unnecessary work on x86-64 and to gain the performance
back, split up vmalloc_sync_all() into two functions:

	* vmalloc_sync_mappings(), and
	* vmalloc_sync_unmappings()

Most call-sites to vmalloc_sync_all() only care about new mappings being
synchronized.  The only exception is the new call-site added in the
above mentioned commit.

Shile Zhang directed us to a report of an 80% regression in reaim
throughput.

Fixes: 3f8fd02b1bf1 ("mm/vmalloc: Sync unmappings in __purge_vmap_area_lazy()")
Reported-by: kernel test robot <oliver.sang@intel.com>
Reported-by: Shile Zhang <shile.zhang@linux.alibaba.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Tested-by: Borislav Petkov <bp@suse.de>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>	[GHES]
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: <stable@vger.kernel.org>
Link: http://lkml.kernel.org/r/20191009124418.8286-1-joro@8bytes.org
Link: https://lists.01.org/hyperkitty/list/lkp@lists.01.org/thread/4D3JPPHBNOSPFK2KEPC6KGKS6J25AIDB/
Link: http://lkml.kernel.org/r/20191113095530.228959-1-shile.zhang@linux.alibaba.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull futex fix from Thomas Gleixner:
 "Fix for yet another subtle futex issue.

  The futex code used ihold() to prevent inodes from vanishing, but
  ihold() does not guarantee inode persistence. Replace the inode
  pointer with a per boot, machine wide, unique inode identifier.

  The second commit fixes the breakage of the hash mechanism which
  causes a 100% performance regression"

* tag 'locking-urgent-2020-03-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  futex: Unbreak futex hashing
  futex: Fix inode life-time issue

Pull timer fix from Thomas Gleixner:
 "A single fix adding the missing time namespace adjustment in
  sys/sysinfo which caused sys/sysinfo to be inconsistent with
  /proc/uptime when read from a task inside a time namespace"

* tag 'timers-urgent-2020-03-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sys/sysinfo: Respect boottime inside time namespace

Pull networking fixes from David Miller:
 "It looks like a decent sized set of fixes, but a lot of these are one
  liner off-by-one and similar type changes:

   1) Fix netlink header pointer to calcular bad attribute offset
      reported to user. From Pablo Neira Ayuso.

   2) Don't double clear PHY interrupts when ->did_interrupt is set,
      from Heiner Kallweit.

   3) Add missing validation of various (devlink, nl802154, fib, etc.)
      attributes, from Jakub Kicinski.

   4) Missing *pos increments in various netfilter seq_next ops, from
      Vasily Averin.

   5) Missing break in of_mdiobus_register() loop, from Dajun Jin.

   6) Don't double bump tx_dropped in veth driver, from Jiang Lidong.

   7) Work around FMAN erratum A050385, from Madalin Bucur.

   8) Make sure ARP header is pulled early enough in bonding driver,
      from Eric Dumazet.

   9) Do a cond_resched() during multicast processing of ipvlan and
      macvlan, from Mahesh Bandewar.

  10) Don't attach cgroups to unrelated sockets when in interrupt
      context, from Shakeel Butt.

  11) Fix tpacket ring state management when encountering unknown GSO
      types. From Willem de Bruijn.

  12) Fix MDIO bus PHY resume by checking mdio_bus_phy_may_suspend()
      only in the suspend context. From Heiner Kallweit"

* git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (112 commits)
  net: systemport: fix index check to avoid an array out of bounds access
  tc-testing: add ETS scheduler to tdc build configuration
  net: phy: fix MDIO bus PM PHY resuming
  net: hns3: clear port base VLAN when unload PF
  net: hns3: fix RMW issue for VLAN filter switch
  net: hns3: fix VF VLAN table entries inconsistent issue
  net: hns3: fix "tc qdisc del" failed issue
  taprio: Fix sending packets without dequeueing them
  net: mvmdio: avoid error message for optional IRQ
  net: dsa: mv88e6xxx: Add missing mask of ATU occupancy register
  net: memcg: fix lockdep splat in inet_csk_accept()
  s390/qeth: implement smarter resizing of the RX buffer pool
  s390/qeth: refactor buffer pool code
  s390/qeth: use page pointers to manage RX buffer pool
  seg6: fix SRv6 L2 tunnels to use IANA-assigned protocol number
  net: dsa: Don't instantiate phylink for CPU/DSA ports unless needed
  net/packet: tpacket_rcv: do not increment ring index on drop
  sxgbe: Fix off by one in samsung driver strncpy size arg
  net: caif: Add lockdep expression to RCU traversal primitive
  MAINTAINERS: remove Sathya Perla as Emulex NIC maintainer
  ...

Pull thread fix from Christian Brauner:
 "This contains a single fix for a regression which was introduced when
  we introduced the ability to select a specific pid at process creation
  time.

  When this feature is requested, the error value will be set to -EPERM
  after exiting the pid allocation loop. This caused EPERM to be
  returned when e.g. the init process/child subreaper of the pid
  namespace has already died where we used to return ENOMEM before.

  The first patch here simply fixes the regression by unconditionally
  setting the return value back to ENOMEM again once we've successfully
  allocated the requested pid number. This should be easy to backport to
  v5.5.

  The second patch adds a comment explaining that we must keep returning
  ENOMEM since we've been doing it for a long time and have explicitly
  documented this behavior for userspace. This seemed worthwhile because
  we now have at least two separate example where people tried to change
  the return value to something other than ENOMEM (The first version of
  the regression fix did that too and the commit message links to an
  earlier patch that tried to do the same.).

  I have a simple regression test to make sure we catch this regression
  in the future but since that introduces a whole new selftest subdir
  and test files I'll keep this for v5.7"

* tag 'for-linus-2020-03-10' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux:
  pid: make ENOMEM return value more obvious
  pid: Fix error return value in some cases

Pull ftrace fix from Steven Rostedt:
 "Have ftrace lookup_rec() return a consistent record otherwise it can
  break live patching"

* tag 'trace-v5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
  ftrace: Return the first found result in lookup_rec()

It appears that ip ranges can overlap so. In that case lookup_rec()
returns whatever results it got last even if it found nothing in last
searched page.

This breaks an obscure livepatch late module patching usecase:
  - load livepatch
  - load the patched module
  - unload livepatch
  - try to load livepatch again

To fix this return from lookup_rec() as soon as it found the record
containing searched-for ip. This used to be this way prior lookup_rec()
introduction.

Link: http://lkml.kernel.org/r/20200306174317.21699-1-asavkov@redhat.com

Cc: stable@vger.kernel.org
Fixes: 7e16f581a817 ("ftrace: Separate out functionality from ftrace_location_range()")
Signed-off-by: Artem Savkov <asavkov@redhat.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>

We are testing network memory accounting in our setup and noticed
inconsistent network memory usage and often unrelated cgroups network
usage correlates with testing workload. On further inspection, it
seems like mem_cgroup_sk_alloc() and cgroup_sk_alloc() are broken in
irq context specially for cgroup v1.

mem_cgroup_sk_alloc() and cgroup_sk_alloc() can be called in irq context
and kind of assumes that this can only happen from sk_clone_lock()
and the source sock object has already associated cgroup. However in
cgroup v1, where network memory accounting is opt-in, the source sock
can be unassociated with any cgroup and the new cloned sock can get
associated with unrelated interrupted cgroup.

Cgroup v2 can also suffer if the source sock object was created by
process in the root cgroup or if sk_alloc() is called in irq context.
The fix is to just do nothing in interrupt.

WARNING: Please note that about half of the TCP sockets are allocated
from the IRQ context, so, memory used by such sockets will not be
accouted by the memcg.

The stack trace of mem_cgroup_sk_alloc() from IRQ-context:

CPU: 70 PID: 12720 Comm: ssh Tainted:  5.6.0-smp-DEV #1
Hardware name: ...
Call Trace:
 <IRQ>
 dump_stack+0x57/0x75
 mem_cgroup_sk_alloc+0xe9/0xf0
 sk_clone_lock+0x2a7/0x420
 inet_csk_clone_lock+0x1b/0x110
 tcp_create_openreq_child+0x23/0x3b0
 tcp_v6_syn_recv_sock+0x88/0x730
 tcp_check_req+0x429/0x560
 tcp_v6_rcv+0x72d/0xa40
 ip6_protocol_deliver_rcu+0xc9/0x400
 ip6_input+0x44/0xd0
 ? ip6_protocol_deliver_rcu+0x400/0x400
 ip6_rcv_finish+0x71/0x80
 ipv6_rcv+0x5b/0xe0
 ? ip6_sublist_rcv+0x2e0/0x2e0
 process_backlog+0x108/0x1e0
 net_rx_action+0x26b/0x460
 __do_softirq+0x104/0x2a6
 do_softirq_own_stack+0x2a/0x40
 </IRQ>
 do_softirq.part.19+0x40/0x50
 __local_bh_enable_ip+0x51/0x60
 ip6_finish_output2+0x23d/0x520
 ? ip6table_mangle_hook+0x55/0x160
 __ip6_finish_output+0xa1/0x100
 ip6_finish_output+0x30/0xd0
 ip6_output+0x73/0x120
 ? __ip6_finish_output+0x100/0x100
 ip6_xmit+0x2e3/0x600
 ? ipv6_anycast_cleanup+0x50/0x50
 ? inet6_csk_route_socket+0x136/0x1e0
 ? skb_free_head+0x1e/0x30
 inet6_csk_xmit+0x95/0xf0
 __tcp_transmit_skb+0x5b4/0xb20
 __tcp_send_ack.part.60+0xa3/0x110
 tcp_send_ack+0x1d/0x20
 tcp_rcv_state_process+0xe64/0xe80
 ? tcp_v6_connect+0x5d1/0x5f0
 tcp_v6_do_rcv+0x1b1/0x3f0
 ? tcp_v6_do_rcv+0x1b1/0x3f0
 __release_sock+0x7f/0xd0
 release_sock+0x30/0xa0
 __inet_stream_connect+0x1c3/0x3b0
 ? prepare_to_wait+0xb0/0xb0
 inet_stream_connect+0x3b/0x60
 __sys_connect+0x101/0x120
 ? __sys_getsockopt+0x11b/0x140
 __x64_sys_connect+0x1a/0x20
 do_syscall_64+0x51/0x200
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

The stack trace of mem_cgroup_sk_alloc() from IRQ-context:
Fixes: 2d7580738345 ("mm: memcontrol: consolidate cgroup socket tracking")
Fixes: d979a39d7242 ("cgroup: duplicate cgroup reference when cloning sockets")
Signed-off-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull cgroup fixes from Tejun Heo:

 - cgroup.procs listing related fixes.

   It didn't interlock properly with exiting tasks leaving a short
   window where a cgroup has empty cgroup.procs but still can't be
   removed and misbehaved on short reads.

 - psi_show() crash fix on 32bit ino archs

 - Empty release_agent handling fix

* 'for-5.6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup1: don't call release_agent when it is ""
  cgroup: fix psi_show() crash on 32bit ino archs
  cgroup: Iterate tasks that did not finish do_exit()
  cgroup: cgroup_procs_next should increase position index
  cgroup-v1: cgroup_pidlist_next should update position index

Pull workqueue fixes from Tejun Heo:
 "Workqueue has been incorrectly round-robining per-cpu work items.
  Hillf's patch fixes that.

  The other patch documents memory-ordering properties of workqueue
  operations"

* 'for-5.6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
  workqueue: don't use wq_select_unbound_cpu() for bound works
  workqueue: Document (some) memory-ordering properties of {queue,schedule}_work()