Daniel Borkmann says:

====================
pull-request: bpf-next 2018-10-21

The following pull-request contains BPF updates for your *net-next* tree.

The main changes are:

1) Implement two new kind of BPF maps, that is, queue and stack
   map along with new peek, push and pop operations, from Mauricio.

2) Add support for MSG_PEEK flag when redirecting into an ingress
   psock sk_msg queue, and add a new helper bpf_msg_push_data() for
   insert data into the message, from John.

3) Allow for BPF programs of type BPF_PROG_TYPE_CGROUP_SKB to use
   direct packet access for __skb_buff, from Song.

4) Use more lightweight barriers for walking perf ring buffer for
   libbpf and perf tool as well. Also, various fixes and improvements
   from verifier side, from Daniel.

5) Add per-symbol visibility for DSO in libbpf and hide by default
   global symbols such as netlink related functions, from Andrey.

6) Two improvements to nfp's BPF offload to check vNIC capabilities
   in case prog is shared with multiple vNICs and to protect against
   mis-initializing atomic counters, from Jakub.

7) Fix for bpftool to use 4 context mode for the nfp disassembler,
   also from Jakub.

8) Fix a return value comparison in test_libbpf.sh and add several
   bpftool improvements in bash completion, documentation of bpf fs
   restrictions and batch mode summary print, from Quentin.

9) Fix a file resource leak in BPF selftest's load_kallsyms()
   helper, from Peng.

10) Fix an unused variable warning in map_lookup_and_delete_elem(),
    from Alexei.

11) Fix bpf_skb_adjust_room() signature in BPF UAPI helper doc,
    from Nicolas.

12) Add missing executables to .gitignore in BPF selftests, from Anders.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Wolfram Sang says:

====================
net: simplify getting .driver_data

I got tired of fixing this in Renesas drivers manually, so I took the big
hammer. Remove this cumbersome code pattern which got copy-pasted too much
already:

-	struct platform_device *pdev = to_platform_device(dev);
-	struct ep93xx_keypad *keypad = platform_get_drvdata(pdev);
+	struct ep93xx_keypad *keypad = dev_get_drvdata(dev);

A branch, tested by buildbot, can be found here:

git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux.git coccinelle/get_drvdata

I have been asked if it couldn't be done for dev_set_drvdata as well. I checked
it and did not find one occasion where it could be simplified like this. Not
much of a surprise because driver_data is usually set in probe() functions
which access struct platform_device in many other ways.

I am open for other comments, suggestions, too, of course.

Here is the cocci-script I created:

@@
struct device* d;
identifier pdev;
expression *ptr;
@@
(
-	struct platform_device *pdev = to_platform_device(d);
|
-	struct platform_device *pdev;
	...
-	pdev = to_platform_device(d);
)
	<... when != pdev
-	&pdev->dev
+	d
	...>

	ptr =
-	platform_get_drvdata(pdev)
+	dev_get_drvdata(d)

	<... when != pdev
-	&pdev->dev
+	d
	...>
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

David Ahern's dump indexing bug fix in 'net' overlapped the
change of the function signature of inet6_fill_ifaddr() in
'net-next'.  Trivially resolved.
Signed-off-by: David S. Miller <davem@davemloft.net>

When trying to complete "bpftool map update" commands, the call to
printf would print an error message that would show on the command line
if no map is found to complete the command line.

Fix it by making sure we have map ids to complete the line with, before
we try to print something.
Signed-off-by: Quentin Monnet <quentin.monnet@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

When batch mode is used and all commands succeeds, bpftool prints the
number of commands processed to stderr. There is no particular reason to
use stderr for this, we could as well use stdout. It would avoid getting
unnecessary output on stderr if the standard ouptut is redirected, for
example.
Reported-by: David Beckett <david.beckett@netronome.com>
Signed-off-by: Quentin Monnet <quentin.monnet@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

Names used to pin eBPF programs and maps under the eBPF virtual file
system cannot contain a dot character, which is reserved for future
extensions of this file system.

Document this in bpftool man pages to avoid users getting confused if
pinning fails because of a dot.
Signed-off-by: Quentin Monnet <quentin.monnet@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

David writes:
  "Networking:

   A few straggler bug fixes:

   1) Fix indexing of multi-pass dumps of ipv6 addresses, from David
      Ahern.

   2) Revert RCU locking change for bonding netpoll, causes worse
      problems than it solves.

   3) pskb_trim_rcsum_slow() doesn't handle odd trim offsets, resulting
      in erroneous bad hw checksum triggers with CHECKSUM_COMPLETE
      devices.  From Dimitris Michailidis.

   4) a revert to some neighbour code changes that adjust notifications
      in a way that confuses some apps."

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
  Revert "neighbour: force neigh_invalidate when NUD_FAILED update is from admin"
  net/ipv6: Fix index counter for unicast addresses in in6_dump_addrs
  net: fix pskb_trim_rcsum_slow() with odd trim offset
  Revert "bond: take rcu lock in netpoll_send_skb_on_dev"

The return value for each test in test_libbpf.sh is compared with

    if (( $? == 0 )) ; then ...

This works well with bash, but not with dash, that /bin/sh is aliased to
on some systems (such as Ubuntu).

Let's replace this comparison by something that works on both shells.
Signed-off-by: Quentin Monnet <quentin.monnet@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

Daniel Borkmann says:

====================
Last batch of misc patches I had in queue: first one removes some left-over
bits from ULP, second is a fix in the verifier where we wrongly use register
number as type to fetch the string for the dump, third disables xadd on flow
keys and subsequent one removes the flow key type from check_helper_mem_access()
as they cannot be passed into any helper as of today. Next one lets map push,
pop, peek avoid having to go through retpoline, and last one has a couple of
minor fixes and cleanups for the ring buffer walk.
====================
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

Simplify bpf_perf_event_read_simple() a bit and fix up some minor
things along the way: the return code in the header is not of type
int but enum bpf_perf_event_ret instead. Once callback indicated
to break the loop walking event data, it also needs to be consumed
in data_tail since it has been processed already.

Moreover, bpf_perf_event_print_t callback should avoid void * as
we actually get a pointer to struct perf_event_header and thus
applications can make use of container_of() to have type checks.
The walk also doesn't have to use modulo op since the ring size is
required to be power of two.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

Extend prior work from 09772d92 ("bpf: avoid retpoline for
lookup/update/delete calls on maps") to also apply to the recently
added map helpers that perform push/pop/peek operations so that
the indirect call can be avoided.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

They PTR_TO_FLOW_KEYS is not used today to be passed into a helper
as memory, so it can be removed from check_helper_mem_access().
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

We should not enable xadd operation for flow key memory if not
needed there anyway. There is no such issue as described in the
commit f37a8cb8 ("bpf: reject stores into ctx via st and xadd")
since there's no context rewriter for flow keys today, but it
also shouldn't become part of the user facing behavior to allow
for it. After patch:

  0: (79) r7 = *(u64 *)(r1 +144)
  1: (b7) r3 = 4096
  2: (db) lock *(u64 *)(r7 +0) += r3
  BPF_XADD stores into R7 flow_keys is not allowed
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

Using reg_type_str[insn->dst_reg] is incorrect since insn->dst_reg
contains the register number but not the actual register type. Add
a small reg_state() helper and use it to get to the type. Also fix
up the test_verifier test cases that have an incorrect errstr.

Fixes: 9d2be44a ("bpf: Reuse canonical string formatter for ctx errs")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

They are not used anymore and therefore should be removed.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

This reverts commit 8e326289.

This patch results in unnecessary netlink notification when one
tries to delete a neigh entry already in NUD_FAILED state. Found
this with a buggy app that tries to delete a NUD_FAILED entry
repeatedly. While the notification issue can be fixed with more
checks, adding more complexity here seems unnecessary. Also,
recent tests with other changes in the neighbour code have
shown that the INCOMPLETE and PROBE checks are good enough for
the original issue.
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The loop wants to skip previously dumped addresses, so loops until
current index >= saved index. If the message fills it wants to save
the index for the next address to dump - ie., the one that did not
fit in the current message.

Currently, it is incrementing the index counter before comparing to the
saved index, and then the saved index is off by 1 - it assumes the
current address is going to fit in the message.

Change the index handling to increment only after a succesful dump.

Fixes: 502a2ffd ("ipv6: convert idev_list to list macros")
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

John Fastabend says:

====================
This series adds a new helper bpf_msg_push_data to be used by
sk_msg programs. The helper can be used to insert extra bytes into
the message that can then be used by the program as metadata tags
among other things.

The first patch adds the helper, second patch the libbpf support,
and last patch updates test_sockmap to run msg_push_data tests.

v2: rebase after queue map and in filter.c convert int -> u32
====================
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

Add options to run msg_push_data, this patch creates two more flags
in test_sockmap that can be used to specify the offset and length
of bytes to be added. The new options are --txmsg_start_push to
specify where bytes should be inserted and --txmsg_end_push to
specify how many bytes. This is analagous to the options that are
used to pull data, --txmsg_start and --txmsg_end.

In addition to adding the options tests are added to the test
suit to run the tests similar to what was done for msg_pull_data.
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

Add support for new bpf_msg_push_data in libbpf.
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

This allows user to push data into a msg using sk_msg program types.
The format is as follows,

	bpf_msg_push_data(msg, offset, len, flags)

this will insert 'len' bytes at offset 'offset'. For example to
prepend 10 bytes at the front of the message the user can,

	bpf_msg_push_data(msg, 0, 10, 0);

This will invalidate data bounds so BPF user will have to then recheck
data bounds after calling this. After this the msg size will have been
updated and the user is free to write into the added bytes. We allow
any offset/len as long as it is within the (data, data_end) range.
However, a copy will be required if the ring is full and its possible
for the helper to fail with ENOMEM or EINVAL errors which need to be
handled by the BPF program.

This can be used similar to XDP metadata to pass data between sk_msg
layer and lower layers.
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

This patch is basically a resubmit of 1e918876 ("r8169: add support
for Byte Queue Limits") which was reverted later. The problems causing
the revert seem to have been fixed in the meantime.
Only change to the original patch is that the call to
netdev_reset_queue was moved to rtl8169_tx_clear.

The Tested-by refers to a system using the RTL8168evl chip version.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Tested-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Having a separate "slow event" handler isn't needed because all
interrupt events trigger asynchronous activity. And in case of SYSErr
we have bigger problems than performance anyway.
This patch also allows to get rid of acking interrupt events in the
NAPI poll callback.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth-next

Johan Hedberg says:

====================
pull request: bluetooth-next 2018-10-20

Here's one more bluetooth-next pull request for the 4.20 kernel.

 - Added new USB ID for QCA_ROME controller
 - Added debug trace support from QCA wcn3990 controllers
 - Updated L2CAP to conform to latest Errata Service Release
 - Fix binding to non-removable BCM43430 devices

Please let me know if there are any issues pulling. Thanks.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Pablo Neira Ayuso says:

====================
Netfilter updates for net-next

The following patchset contains Netfilter updates for your net-next tree:

1) Use lockdep_is_held() in ipset_dereference_protected(), from Lance Roy.

2) Remove unused variable in cttimeout, from YueHaibing.

3) Add ttl option for nft_osf, from Fernando Fernandez Mancera.

4) Use xfrm family to deal with IPv6-in-IPv4 packets from nft_xfrm,
   from Florian Westphal.

5) Simplify xt_osf_match_packet().

6) Missing ct helper alias definition in snmp_trap helper, from Taehee Yoo.

7) Remove unnecessary parameter in nf_flow_table_cleanup(), from Taehee Yoo.

8) Remove unused variable definitions in nft_{dup,fwd}, from Weongyo Jeong.

9) Remove empty net/netfilter/nfnetlink_log.h file, from Taehee Yoo.

10) Revert xt_quota updates remain option due to problems in the listing
    path for 32-bit arches, from Maze.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Ingo writes:
  "x86 fixes:

   It's 4 misc fixes, 3 build warning fixes and 3 comment fixes.

   In hindsight I'd have left out the 3 comment fixes to make the pull
   request look less scary at such a late point in the cycle. :-/"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/swiotlb: Enable swiotlb for > 4GiG RAM on 32-bit kernels
  x86/fpu: Fix i486 + no387 boot crash by only saving FPU registers on context switch if there is an FPU
  x86/fpu: Remove second definition of fpu in __fpu__restore_sig()
  x86/entry/64: Further improve paranoid_entry comments
  x86/entry/32: Clear the CS high bits
  x86/boot: Add -Wno-pointer-sign to KBUILD_CFLAGS
  x86/time: Correct the attribute on jiffies' definition
  x86/entry: Add some paranoid entry/exit CR3 handling comments
  x86/percpu: Fix this_cpu_read()
  x86/tsc: Force inlining of cyc2ns bits

Ingo writes:
  "scheduler fixes:

   Two fixes: a CFS-throttling bug fix, and an interactivity fix."

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/fair: Fix the min_vruntime update logic in dequeue_entity()
  sched/fair: Fix throttle_list starvation with low CFS quota

Ingo writes:
  "perf fixes:

   Misc perf tooling fixes."

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf tools: Stop fallbacking to kallsyms for vdso symbols lookup
  perf tools: Pass build flags to traceevent build
  perf report: Don't crash on invalid inline debug information
  perf cpu_map: Align cpu map synthesized events properly.
  perf tools: Fix tracing_path_mount proper path
  perf tools: Fix use of alternatives to find JDIR
  perf evsel: Store ids for events with their own cpus perf_event__synthesize_event_update_cpus
  perf vendor events intel: Fix wrong filter_band* values for uncore events
  Revert "perf tools: Fix PMU term format max value calculation"
  tools headers uapi: Sync kvm.h copy
  tools arch uapi: Sync the x86 kvm.h copy

We've been getting checksum errors involving small UDP packets, usually
59B packets with 1 extra non-zero padding byte. netdev_rx_csum_fault()
has been complaining that HW is providing bad checksums. Turns out the
problem is in pskb_trim_rcsum_slow(), introduced in commit 88078d98
("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends").

The source of the problem is that when the bytes we are trimming start
at an odd address, as in the case of the 1 padding byte above,
skb_checksum() returns a byte-swapped value. We cannot just combine this
with skb->csum using csum_sub(). We need to use csum_block_sub() here
that takes into account the parity of the start address and handles the
swapping.

Matches existing code in __skb_postpull_rcsum() and esp_remove_trailer().

Fixes: 88078d98 ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
Signed-off-by: Dimitris Michailidis <dmichail@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

At least UDP / TCP stacks can now cook skbs with a tstamp using
MONOTONIC base (or arbitrary values with SCM_TXTIME)

Since loopback driver does not call (directly or indirectly)
skb_scrub_packet(), we need to clear skb->tstamp so that
net_timestamp_check() can eventually resample the time,
using ktime_get_real().

Fixes: 80b14dee ("net: Add a new socket option for a future transmit time.")
Fixes: fb420d5d ("tcp/fq: move back to CLOCK_MONOTONIC")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Cc: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Dave writes:
  "drm fixes for 4.19 final (part 2)

   Looked like two stragglers snuck in, one very urgent the pageflipping
   was missing a reference that could result in a GPF on non-i915
   drivers, the other is an overflow in the sun4i dotclock calcs
   resulting in a mode not getting set."

* tag 'drm-fixes-2018-10-20-1' of git://anongit.freedesktop.org/drm/drm:
  drm/sun4i: Fix an ulong overflow in the dotclock driver
  drm: Get ref on CRTC commit object when waiting for flip_done