Dean Jenkins says:

====================
asix: Improve robustness

Please consider taking these patches to improve the robustness of the ASIX USB
to Ethernet driver.

Failures prompting an ASIX driver code review
=============================================

On an ARM i.MX6 embedded platform some strange one-off and two-off failures were
observed in and around the ASIX USB to Ethernet driver. This was observed on a
highly modified kernel 3.14 with the ASIX driver containing back-ported changes
from kernel.org up to kernel 4.8 approximately.

a) A one-off failure in asix_rx_fixup_internal():

There was an occurrence of an attempt to write off the end of the netdev buffer
which was trapped by skb_over_panic() in skb_put().

[20030.846440] skbuff: skb_over_panic: text:7f2271c0 len:120 put:60 head:8366ecc0 data:8366ed02 tail:0x8366ed7a end:0x8366ed40 dev:eth0
[20030.863007] Kernel BUG at 8044ce38 [verbose debug info unavailable]

[20031.215345] Backtrace:
[20031.217884] [<8044cde0>] (skb_panic) from [<8044d50c>] (skb_put+0x50/0x5c)
[20031.227408] [<8044d4bc>] (skb_put) from [<7f2271c0>] (asix_rx_fixup_internal+0x1c4/0x23c [asix])
[20031.242024] [<7f226ffc>] (asix_rx_fixup_internal [asix]) from [<7f22724c>] (asix_rx_fixup_common+0x14/0x18 [asix])
[20031.260309] [<7f227238>] (asix_rx_fixup_common [asix]) from [<7f21f7d4>] (usbnet_bh+0x74/0x224 [usbnet])
[20031.269879] [<7f21f760>] (usbnet_bh [usbnet]) from [<8002f834>] (call_timer_fn+0xa4/0x1f0)
[20031.283961] [<8002f790>] (call_timer_fn) from [<80030834>] (run_timer_softirq+0x230/0x2a8)
[20031.302782] [<80030604>] (run_timer_softirq) from [<80028780>] (__do_softirq+0x15c/0x37c)
[20031.321511] [<80028624>] (__do_softirq) from [<80028c38>] (irq_exit+0x8c/0xe8)
[20031.339298] [<80028bac>] (irq_exit) from [<8000e9c8>] (handle_IRQ+0x8c/0xc8)
[20031.350038] [<8000e93c>] (handle_IRQ) from [<800085c8>] (gic_handle_irq+0xb8/0xf8)
[20031.365528] [<80008510>] (gic_handle_irq) from [<8050de80>] (__irq_svc+0x40/0x70)

Analysis of the logic of the ASIX driver (containing backported changes from
kernel.org up to kernel 4.8 approximately) suggested that the software could not
trigger skb_over_panic(). The analysis of the kernel BUG() crash information
suggested that the netdev buffer was written with 2 minimal 60 octet length
Ethernet frames (ASIX hardware drops the 4 octet FCS field) and the 2nd Ethernet
frame attempted to write off the end of the netdev buffer.

Note that the netdev buffer should only contain 1 Ethernet frame so if an
attempt to write 2 Ethernet frames into the buffer is made then that is wrong.
However, the logic of the asix_rx_fixup_internal() only allows 1 Ethernet frame
to be written into the netdev buffer.

Potentially this failure was due to memory corruption because it was only seen
once.

b) Two-off failures in the NAPI layer's backlog queue:

There were 2 crashes in the NAPI layer's backlog queue presumably after
asix_rx_fixup_internal() called usbnet_skb_return().

[24097.273945] Unable to handle kernel NULL pointer dereference at virtual address 00000004

[24097.398944] PC is at process_backlog+0x80/0x16c

[24097.569466] Backtrace:
[24097.572007] [<8045ad98>] (process_backlog) from [<8045b64c>] (net_rx_action+0xcc/0x248)
[24097.591631] [<8045b580>] (net_rx_action) from [<80028780>] (__do_softirq+0x15c/0x37c)
[24097.610022] [<80028624>] (__do_softirq) from [<800289cc>] (run_ksoftirqd+0x2c/0x84)

and

[ 1059.828452] Unable to handle kernel NULL pointer dereference at virtual address 00000000

[ 1059.953715] PC is at process_backlog+0x84/0x16c

[ 1060.140896] Backtrace:
[ 1060.143434] [<8045ad98>] (process_backlog) from [<8045b64c>] (net_rx_action+0xcc/0x248)
[ 1060.163075] [<8045b580>] (net_rx_action) from [<80028780>] (__do_softirq+0x15c/0x37c)
[ 1060.181474] [<80028624>] (__do_softirq) from [<80028c38>] (irq_exit+0x8c/0xe8)
[ 1060.199256] [<80028bac>] (irq_exit) from [<8000e9c8>] (handle_IRQ+0x8c/0xc8)
[ 1060.210006] [<8000e93c>] (handle_IRQ) from [<800085c8>] (gic_handle_irq+0xb8/0xf8)
[ 1060.225492] [<80008510>] (gic_handle_irq) from [<8050de80>] (__irq_svc+0x40/0x70)

The embedded board was only using an ASIX USB to Ethernet adaptor eth0.

Analysis suggested that the doubly-linked list pointers of the backlog queue had
been corrupted because one of the link pointers was NULL.

Potentially this failure was due to memory corruption because it was only seen
twice.

Results of the ASIX driver code review
======================================

During the code review some weaknesses were observed in the ASIX driver and the
following patches have been created to improve the robustness.

Brief overview of the patches
-----------------------------

1. asix: Add rx->ax_skb = NULL after usbnet_skb_return()

The current ASIX driver sends the received Ethernet frame to the NAPI layer of
the network stack via the call to usbnet_skb_return() in
asix_rx_fixup_internal() but retains the rx->ax_skb pointer to the netdev
buffer. The driver no longer needs the rx->ax_skb pointer at this point because
the NAPI layer now has the Ethernet frame.

This means that asix_rx_fixup_internal() must not use rx->ax_skb after the call
to usbnet_skb_return() because it could corrupt the handling of the Ethernet
frame within the network layer.

Therefore, to remove the risk of erroneous usage of rx->ax_skb, set rx->ax_skb
to NULL after the call to usbnet_skb_return(). This avoids potential erroneous
freeing of rx->ax_skb and erroneous writing to the netdev buffer.  If the
software now somehow inappropriately reused rx->ax_skb, then a NULL pointer
dereference of rx->ax_skb would occur which makes investigation easier.

2. asix: Ensure asix_rx_fixup_info members are all reset

This patch creates reset_asix_rx_fixup_info() to allow all the
asix_rx_fixup_info structure members to be consistently reset to initial
conditions.

Call reset_asix_rx_fixup_info() upon each detectable error condition so that the
next URB is processed from a known state.

Otherwise, there is a risk that some members of the asix_rx_fixup_info structure
may be incorrect after an error occurred so potentially leading to a
malfunction.

3. asix: Fix small memory leak in ax88772_unbind()

This patch creates asix_rx_fixup_common_free() to allow the rx->ax_skb to be
freed when necessary.

asix_rx_fixup_common_free() is called from ax88772_unbind() before the parent
private data structure is freed.

Without this patch, there is a risk of a small netdev buffer memory leak each
time ax88772_unbind() is called during the reception of an Ethernet frame that
spans across 2 URBs.

Testing
=======

The patches have been sanity tested on a 64-bit Linux laptop running kernel
4.13-rc2 with the 3 patches applied on top.

The ASIX USB to Adaptor used for testing was (output of lsusb):
ID 0b95:772b ASIX Electronics Corp. AX88772B

Test #1
-------

The test ran a flood ping test script which slowly incremented the ICMP Echo
Request's payload from 0 to 5000 octets. This eventually causes IPv4
fragmentation to occur which causes Ethernet frames to be sent very close to
each other so increases the probability that an Ethernet frame will span 2 URBs.
The test showed that all pings were successful. The test took about 15 minutes
to complete.

Test #2
-------

A script was run on the laptop to periodically run ifdown and ifup every second
so that the ASIX USB to Adaptor was up for 1 second and down for 1 second.

From a Linux PC connected to the laptop, the following ping command was used
ping -f -s 5000 <ip address of laptop>

The large ICMP payload causes IPv4 fragmentation resulting in multiple
Ethernet frames per original IP packet.

Kernel debug within the ASIX driver was enabled to see whether any ASIX errors
were generated. The test was run for about 24 hours and no ASIX errors were
seen.

Patches
=======

The 3 patches have been rebased off the net-next repo master branch with HEAD
fbbeefdd net: fec: Allow reception of frames bigger than 1522 bytes
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

When Ethernet frames span mulitple URBs, the netdev buffer memory
pointed to by the asix_rx_fixup_info structure remains allocated
during the time gap between the 2 executions of asix_rx_fixup_internal().

This means that if ax88772_unbind() is called within this time
gap to free the memory of the parent private data structure then
a memory leak of the part filled netdev buffer memory will occur.

Therefore, create a new function asix_rx_fixup_common_free() to
free the memory of the netdev buffer and add a call to
asix_rx_fixup_common_free() from inside ax88772_unbind().

Consequently when an unbind occurs part way through receiving
an Ethernet frame, the netdev buffer memory that is holding part
of the received Ethernet frame will now be freed.
Signed-off-by: Dean Jenkins <Dean_Jenkins@mentor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There is a risk that the members of the structure asix_rx_fixup_info
become unsynchronised leading to the possibility of a malfunction.

For example, rx->split_head was not being set to false after an
error was detected so potentially could cause a malformed 32-bit
Data header word to be formed.

Therefore add function reset_asix_rx_fixup_info() to reset all the
members of asix_rx_fixup_info so that future processing will start
with known initial conditions.

Also, if (skb->len != offset) becomes true then call
reset_asix_rx_fixup_info() so that the processing of the next URB
starts with known initial conditions. Without the call, the check
does nothing which potentially could lead to a malfunction
when the next URB is processed.

In addition, for robustness, call reset_asix_rx_fixup_info() before
every error path's "return 0". This ensures that the next URB is
processed from known initial conditions.
Signed-off-by: Dean Jenkins <Dean_Jenkins@mentor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In asix_rx_fixup_internal() there is a risk that rx->ax_skb gets
reused after passing the Ethernet frame into the network stack via
usbnet_skb_return().

The risks include:

a) asynchronously freeing rx->ax_skb after passing the netdev buffer
   to the NAPI layer which might corrupt the backlog queue.

b) erroneously reusing rx->ax_skb such as calling skb_put_data() multiple
   times which causes writing off the end of the netdev buffer.

Therefore add a defensive rx->ax_skb = NULL after usbnet_skb_return()
so that it is not possible to free rx->ax_skb or to apply
skb_put_data() too many times.
Signed-off-by: Dean Jenkins <Dean_Jenkins@mentor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit 18f3d6be ("selftests/bpf: Add test cases to test narrower ctx field loads")
introduced new eBPF test cases. One of them (test_pkt_md_access.c)
fails on s390x. The BPF verifier error message is:

[root@s8360046 bpf]# ./test_progs
test_pkt_access:PASS:ipv4 349 nsec
test_pkt_access:PASS:ipv6 212 nsec
[....]
libbpf: load bpf program failed: Permission denied
libbpf: -- BEGIN DUMP LOG ---
libbpf:
0: (71) r2 = *(u8 *)(r1 +0)
invalid bpf_context access off=0 size=1

libbpf: -- END LOG --
libbpf: failed to load program 'test1'
libbpf: failed to load object './test_pkt_md_access.o'
Summary: 29 PASSED, 1 FAILED
[root@s8360046 bpf]#

This is caused by a byte endianness issue. S390x is a big endian
architecture.  Pointer access to the lowest byte or halfword of a
four byte value need to add an offset.
On little endian architectures this offset is not needed.

Fix this and use the same approach as the originator used for other files
(for example test_verifier.c) in his original commit.

With this fix the test program test_progs succeeds on s390x:
[root@s8360046 bpf]# ./test_progs
test_pkt_access:PASS:ipv4 236 nsec
test_pkt_access:PASS:ipv6 217 nsec
test_xdp:PASS:ipv4 3624 nsec
test_xdp:PASS:ipv6 1722 nsec
test_l4lb:PASS:ipv4 926 nsec
test_l4lb:PASS:ipv6 1322 nsec
test_tcp_estats:PASS: 0 nsec
test_bpf_obj_id:PASS:get-fd-by-notexist-prog-id 0 nsec
test_bpf_obj_id:PASS:get-fd-by-notexist-map-id 0 nsec
test_bpf_obj_id:PASS:get-prog-info(fd) 0 nsec
test_bpf_obj_id:PASS:get-map-info(fd) 0 nsec
test_bpf_obj_id:PASS:get-prog-info(fd) 0 nsec
test_bpf_obj_id:PASS:get-map-info(fd) 0 nsec
test_bpf_obj_id:PASS:get-prog-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:get-prog-info(next_id->fd) 0 nsec
test_bpf_obj_id:PASS:get-prog-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:get-prog-info(next_id->fd) 0 nsec
test_bpf_obj_id:PASS:check total prog id found by get_next_id 0 nsec
test_bpf_obj_id:PASS:get-map-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:get-map-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:get-map-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:get-map-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:get-map-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:get-map-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:get-map-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:check get-map-info(next_id->fd) 0 nsec
test_bpf_obj_id:PASS:get-map-fd(next_id) 0 nsec
test_bpf_obj_id:PASS:check get-map-info(next_id->fd) 0 nsec
test_bpf_obj_id:PASS:check total map id found by get_next_id 0 nsec
test_pkt_md_access:PASS: 277 nsec
Summary: 30 PASSED, 0 FAILED
[root@s8360046 bpf]#

Fixes: 18f3d6be ("selftests/bpf: Add test cases to test narrower ctx field loads")
Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

The existing sub channel code did not wait for all the sub-channels
to completely initialize. This could lead to race causing crash
in napi_netif_del() from bad list. The existing code would send
an init message, then wait only for the initial response that
the init message was received. It thought it was waiting for
sub channels but really the init response did the wakeup.

The new code keeps track of the number of open channels and
waits until that many are open.

Other issues here were:
  * host might return less sub-channels than was requested.
  * the new init status is not valid until after init was completed.

Fixes: b3e6b82a ("hv_netvsc: Wait for sub-channels to be processed during probe")
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We really must check with #if __BYTE_ORDER == XYZ instead of
just presence of #ifdef __LITTLE_ENDIAN. I noticed that when
actually running this on big endian machine, the latter test
resolves to true for user space, same for #ifdef __BIG_ENDIAN.

E.g., looking at endian.h from libc, both are also defined
there, so we really must test this against __BYTE_ORDER instead
for proper insns selection. For the kernel, such checks are
fine though e.g. see 13da9e20 ("Revert "endian: #define
__BYTE_ORDER"") and 415586c9 ("UAPI: fix endianness conditionals
in M32R's asm/stat.h") for some more context, but not for
user space. Lets also make sure to properly include endian.h.
After that, suite passes for me:

./test_verifier: ELF 64-bit MSB executable, [...]

Linux foo 4.13.0-rc3+ #4 SMP Fri Aug 4 06:59:30 EDT 2017 s390x s390x s390x GNU/Linux

Before fix: Summary: 505 PASSED, 11 FAILED
After  fix: Summary: 516 PASSED,  0 FAILED

Fixes: 18f3d6be ("selftests/bpf: Add test cases to test narrower ctx field loads")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong <yhs@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Even the driver doesn't do anything with the clk source for SGMII
ports it needs to be enabled by doing a devm_clk_get(), if there is
a clk source in DT.

Fixes: 0db01097 ('xgene: Don't fail probe, if there is no clk resource for SGMII interfaces')
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
Tested-by: Laura Abbott <labbott@redhat.com>
Acked-by: Iyappan Subramanian <isubramanian@apm.com>
Tested-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Inexplicably, commit f381bf6d ("MIPS: Add support for eBPF JIT.")
lost a file somewhere on its path to Linus' tree.  Add back the
missing ebpf_jit.c so that we can build with CONFIG_BPF_JIT selected.

This version of ebpf_jit.c is identical to the original except for two
minor change need to resolve conflicts with changes merged from the
BPF branch:

A) Set prog->jited_len = image_size;
B) Use BPF_TAIL_CALL instead of BPF_CALL | BPF_X

Fixes: f381bf6d ("MIPS: Add support for eBPF JIT.")
Signed-off-by: David Daney <david.daney@cavium.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
Two BPF fixes for s390

Found while testing some other work touching JITs.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

The BPF feature test as well as libbpf is missing the __NR_bpf
define for s390 and currently refuses to compile (selftest suite
depends on libbpf as well). Similar issue was fixed some time
ago via b0c47807 ("bpf: Add sparc support to tools and
samples."), just do the same and add definitions.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

While testing some other work that required JIT modifications, I
run into test_bpf causing a hang when JIT enabled on s390. The
problematic test case was the one from ddc665a4 (bpf, arm64:
fix jit branch offset related to ldimm64), and turns out that we
do have a similar issue on s390 as well. In bpf_jit_prog() we
update next instruction address after returning from bpf_jit_insn()
with an insn_count. bpf_jit_insn() returns either -1 in case of
error (e.g. unsupported insn), 1 or 2. The latter is only the
case for ldimm64 due to spanning 2 insns, however, next address
is only set to i + 1 not taking actual insn_count into account,
thus fix is to use insn_count instead of 1. bpf_jit_enable in
mode 2 provides also disasm on s390:

Before fix:

  000003ff800349b6: a7f40003   brc     15,3ff800349bc                 ; target
  000003ff800349ba: 0000               unknown
  000003ff800349bc: e3b0f0700024       stg     %r11,112(%r15)
  000003ff800349c2: e3e0f0880024       stg     %r14,136(%r15)
  000003ff800349c8: 0db0               basr    %r11,%r0
  000003ff800349ca: c0ef00000000       llilf   %r14,0
  000003ff800349d0: e320b0360004       lg      %r2,54(%r11)
  000003ff800349d6: e330b03e0004       lg      %r3,62(%r11)
  000003ff800349dc: ec23ffeda065       clgrj   %r2,%r3,10,3ff800349b6 ; jmp
  000003ff800349e2: e3e0b0460004       lg      %r14,70(%r11)
  000003ff800349e8: e3e0b04e0004       lg      %r14,78(%r11)
  000003ff800349ee: b904002e   lgr     %r2,%r14
  000003ff800349f2: e3b0f0700004       lg      %r11,112(%r15)
  000003ff800349f8: e3e0f0880004       lg      %r14,136(%r15)
  000003ff800349fe: 07fe               bcr     15,%r14

After fix:

  000003ff80ef3db4: a7f40003   brc     15,3ff80ef3dba
  000003ff80ef3db8: 0000               unknown
  000003ff80ef3dba: e3b0f0700024       stg     %r11,112(%r15)
  000003ff80ef3dc0: e3e0f0880024       stg     %r14,136(%r15)
  000003ff80ef3dc6: 0db0               basr    %r11,%r0
  000003ff80ef3dc8: c0ef00000000       llilf   %r14,0
  000003ff80ef3dce: e320b0360004       lg      %r2,54(%r11)
  000003ff80ef3dd4: e330b03e0004       lg      %r3,62(%r11)
  000003ff80ef3dda: ec230006a065       clgrj   %r2,%r3,10,3ff80ef3de6 ; jmp
  000003ff80ef3de0: e3e0b0460004       lg      %r14,70(%r11)
  000003ff80ef3de6: e3e0b04e0004       lg      %r14,78(%r11)          ; target
  000003ff80ef3dec: b904002e   lgr     %r2,%r14
  000003ff80ef3df0: e3b0f0700004       lg      %r11,112(%r15)
  000003ff80ef3df6: e3e0f0880004       lg      %r14,136(%r15)
  000003ff80ef3dfc: 07fe               bcr     15,%r14

test_bpf.ko suite runs fine after the fix.

Fixes: 05462310 ("s390/bpf: Add s390x eBPF JIT compiler backend")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jiri Pirko says:

====================
mlxsw: Couple of fixes

Ido says:

The first patch prevents us from warning about valid situations that can
happen due to the fact that some operations in switchdev are deferred.

Second patch fixes a long standing problem in which we didn't correctly
free resources upon module removal, resulting in a memory leak.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Each multicast group (MID) stores a bitmap of ports to which a packet
should be forwarded to in case an MDB entry associated with the MID is
hit.

Since the initial introduction of IGMP snooping in commit 3a49b4fd
("mlxsw: Adding layer 2 multicast support") the driver didn't correctly
free these multicast groups upon ungraceful situations such as the
removal of the upper bridge device or module removal.

The correct way to fix this is to associate each MID with the bridge
ports member in it and then drop the reference in case the bridge port
is destroyed, but this will result in a lot more code and will be fixed
in net-next.

For now, upon module removal, traverse the MID list and release each
one.

Fixes: 3a49b4fd ("mlxsw: Adding layer 2 multicast support")
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>

Some operations in the bridge driver such as MDB deletion are preformed
in an atomic context and thus deferred to a process context by the
switchdev infrastructure.

Therefore, by the time the operation is performed by the underlying
device driver it's possible the bridge port context is already gone.
This is especially true for removal flows, but theoretically can also be
invoked during addition.

Remove the warnings in such situations and return normally.

Fixes: c57529e1 ("mlxsw: spectrum: Replace vPorts with Port-VLAN")
Fixes: 3922285d ("net: bridge: Add support for offloading port attributes")
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>

Neal Cardwell says:

====================
tcp: fix xmit timer rearming to avoid stalls

This patch series is a bug fix for a TCP loss recovery performance bug
reported independently in recent netdev threads:

 (i)  July 26, 2017: netdev thread "TCP fast retransmit issues"
 (ii) July 26, 2017: netdev thread:
       "[PATCH V2 net-next] TLP: Don't reschedule PTO when there's one
       outstanding TLP retransmission"

Many thanks to Klavs Klavsen and Mao Wenan for the detailed reports,
traces, and packetdrill test cases, which enabled us to root-cause
this issue and verify the fix.

- v1 -> v2:
 - In patch 2/3, changed an unclear comment in the pre-existing code
   in tcp_schedule_loss_probe() to be more clear (thanks to Eric Dumazet
   for suggesting we improve this).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Fix a TCP loss recovery performance bug raised recently on the netdev
list, in two threads:

(i)  July 26, 2017: netdev thread "TCP fast retransmit issues"
(ii) July 26, 2017: netdev thread:
     "[PATCH V2 net-next] TLP: Don't reschedule PTO when there's one
     outstanding TLP retransmission"

The basic problem is that incoming TCP packets that did not indicate
forward progress could cause the xmit timer (TLP or RTO) to be rearmed
and pushed back in time. In certain corner cases this could result in
the following problems noted in these threads:

 - Repeated ACKs coming in with bogus SACKs corrupted by middleboxes
   could cause TCP to repeatedly schedule TLPs forever. We kept
   sending TLPs after every ~200ms, which elicited bogus SACKs, which
   caused more TLPs, ad infinitum; we never fired an RTO to fill in
   the holes.

 - Incoming data segments could, in some cases, cause us to reschedule
   our RTO or TLP timer further out in time, for no good reason. This
   could cause repeated inbound data to result in stalls in outbound
   data, in the presence of packet loss.

This commit fixes these bugs by changing the TLP and RTO ACK
processing to:

 (a) Only reschedule the xmit timer once per ACK.

 (b) Only reschedule the xmit timer if tcp_clean_rtx_queue() deems the
     ACK indicates sufficient forward progress (a packet was
     cumulatively ACKed, or we got a SACK for a packet that was sent
     before the most recent retransmit of the write queue head).

This brings us back into closer compliance with the RFCs, since, as
the comment for tcp_rearm_rto() notes, we should only restart the RTO
timer after forward progress on the connection. Previously we were
restarting the xmit timer even in these cases where there was no
forward progress.

As a side benefit, this commit simplifies and speeds up the TCP timer
arming logic. We had been calling inet_csk_reset_xmit_timer() three
times on normal ACKs that cumulatively acknowledged some data:

1) Once near the top of tcp_ack() to switch from TLP timer to RTO:
        if (icsk->icsk_pending == ICSK_TIME_LOSS_PROBE)
               tcp_rearm_rto(sk);

2) Once in tcp_clean_rtx_queue(), to update the RTO:
        if (flag & FLAG_ACKED) {
               tcp_rearm_rto(sk);

3) Once in tcp_ack() after tcp_fastretrans_alert() to switch from RTO
   to TLP:
        if (icsk->icsk_pending == ICSK_TIME_RETRANS)
               tcp_schedule_loss_probe(sk);

This commit, by only rescheduling the xmit timer once per ACK,
simplifies the code and reduces CPU overhead.

This commit was tested in an A/B test with Google web server
traffic. SNMP stats and request latency metrics were within noise
levels, substantiating that for normal web traffic patterns this is a
rare issue. This commit was also tested with packetdrill tests to
verify that it fixes the timer behavior in the corner cases discussed
in the netdev threads mentioned above.

This patch is a bug fix patch intended to be queued for -stable
relases.

Fixes: 6ba8a3b1 ("tcp: Tail loss probe (TLP)")
Reported-by: Klavs Klavsen <kl@vsen.dk>
Reported-by: Mao Wenan <maowenan@huawei.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Have tcp_schedule_loss_probe() base the TLP scheduling decision based
on when the RTO *should* fire. This is to enable the upcoming xmit
timer fix in this series, where tcp_schedule_loss_probe() cannot
assume that the last timer installed was an RTO timer (because we are
no longer doing the "rearm RTO, rearm RTO, rearm TLP" dance on every
ACK). So tcp_schedule_loss_probe() must independently figure out when
an RTO would want to fire.

In the new TLP implementation following in this series, we cannot
assume that icsk_timeout was set based on an RTO; after processing a
cumulative ACK the icsk_timeout we see can be from a previous TLP or
RTO. So we need to independently recalculate the RTO time (instead of
reading it out of icsk_timeout). Removing this dependency on the
nature of icsk_timeout makes things a little easier to reason about
anyway.

Note that the old and new code should be equivalent, since they are
both saying: "if the RTO is in the future, but at an earlier time than
the normal TLP time, then set the TLP timer to fire when the RTO would
have fired".

Fixes: 6ba8a3b1 ("tcp: Tail loss probe (TLP)")
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pure refactor. This helper will be required in the xmit timer fix
later in the patch series. (Because the TLP logic will want to make
this calculation.)

Fixes: 6ba8a3b1 ("tcp: Tail loss probe (TLP)")
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

After commit c2ed1880 ("net: ipv6: check route protocol when
deleting routes"), ipv6 route checks rt protocol when trying to
remove a rt entry.

It introduced a side effect causing 'ip -6 route flush cache' not
to work well. When flushing caches with iproute, all route caches
get dumped from kernel then removed one by one by sending DELROUTE
requests to kernel for each cache.

The thing is iproute sends the request with the cache whose proto
is set with RTPROT_REDIRECT by rt6_fill_node() when kernel dumps
it. But in kernel the rt_cache protocol is still 0, which causes
the cache not to be matched and removed.

So the real reason is rt6i_protocol in the route is not set when
it is allocated. As David Ahern's suggestion, this patch is to
set rt6i_protocol properly in the route when it is installed and
remove the codes setting rtm_protocol according to rt6i_flags in
rt6_fill_node.

This is also an improvement to keep rt6i_protocol consistent with
rtm_protocol.

Fixes: c2ed1880 ("net: ipv6: check route protocol when deleting routes")
Reported-by: Jianlin Shi <jishi@redhat.com>
Suggested-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

syzkaller was able to trigger a divide by 0 in TCP stack [1]

Issue here is that keepalive timer needs to be updated to not attempt
to send a probe if the connection setup was deferred using
TCP_FASTOPEN_CONNECT socket option added in linux-4.11

[1]
 divide error: 0000 [#1] SMP
 CPU: 18 PID: 0 Comm: swapper/18 Not tainted
 task: ffff986f62f4b040 ti: ffff986f62fa2000 task.ti: ffff986f62fa2000
 RIP: 0010:[<ffffffff8409cc0d>]  [<ffffffff8409cc0d>] __tcp_select_window+0x8d/0x160
 Call Trace:
  <IRQ>
  [<ffffffff8409d951>] tcp_transmit_skb+0x11/0x20
  [<ffffffff8409da21>] tcp_xmit_probe_skb+0xc1/0xe0
  [<ffffffff840a0ee8>] tcp_write_wakeup+0x68/0x160
  [<ffffffff840a151b>] tcp_keepalive_timer+0x17b/0x230
  [<ffffffff83b3f799>] call_timer_fn+0x39/0xf0
  [<ffffffff83b40797>] run_timer_softirq+0x1d7/0x280
  [<ffffffff83a04ddb>] __do_softirq+0xcb/0x257
  [<ffffffff83ae03ac>] irq_exit+0x9c/0xb0
  [<ffffffff83a04c1a>] smp_apic_timer_interrupt+0x6a/0x80
  [<ffffffff83a03eaf>] apic_timer_interrupt+0x7f/0x90
  <EOI>
  [<ffffffff83fed2ea>] ? cpuidle_enter_state+0x13a/0x3b0
  [<ffffffff83fed2cd>] ? cpuidle_enter_state+0x11d/0x3b0

Tested:

Following packetdrill no longer crashes the kernel

`echo 0 >/proc/sys/net/ipv4/tcp_timestamps`

// Cache warmup: send a Fast Open cookie request
    0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3
   +0 fcntl(3, F_SETFL, O_RDWR|O_NONBLOCK) = 0
   +0 setsockopt(3, SOL_TCP, TCP_FASTOPEN_CONNECT, [1], 4) = 0
   +0 connect(3, ..., ...) = -1 EINPROGRESS (Operation is now in progress)
   +0 > S 0:0(0) <mss 1460,nop,nop,sackOK,nop,wscale 8,FO,nop,nop>
 +.01 < S. 123:123(0) ack 1 win 14600 <mss 1460,nop,nop,sackOK,nop,wscale 6,FO abcd1234,nop,nop>
   +0 > . 1:1(0) ack 1
   +0 close(3) = 0
   +0 > F. 1:1(0) ack 1
   +0 < F. 1:1(0) ack 2 win 92
   +0 > .  2:2(0) ack 2

   +0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 4
   +0 fcntl(4, F_SETFL, O_RDWR|O_NONBLOCK) = 0
   +0 setsockopt(4, SOL_TCP, TCP_FASTOPEN_CONNECT, [1], 4) = 0
   +0 setsockopt(4, SOL_SOCKET, SO_KEEPALIVE, [1], 4) = 0
 +.01 connect(4, ..., ...) = 0
   +0 setsockopt(4, SOL_TCP, TCP_KEEPIDLE, [5], 4) = 0
   +10 close(4) = 0

`echo 1 >/proc/sys/net/ipv4/tcp_timestamps`

Fixes: 19f6d3f3 ("net/tcp-fastopen: Add new API support")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Wei Wang <weiwan@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Simon Wunderlich says:

====================
Here is a batman-adv bugfix:

 - fix TT sync flag inconsistency problems, which can lead to excess packets,
   by Linus Luessing
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

If the sender switches the congestion control during ECN-triggered
cwnd-reduction state (CA_CWR), upon exiting recovery cwnd is set to
the ssthresh value calculated by the previous congestion control. If
the previous congestion control is BBR that always keep ssthresh
to TCP_INIFINITE_SSTHRESH, cwnd ends up being infinite. The safe
step is to avoid assigning invalid ssthresh value when recovery ends.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

SCRQ resources are freed during renegotiation, but they are not
re-allocated afterwards due to some changes in the initialization
process. Fix that by re-allocating the memory after renegotation.

SCRQ's can also be freed if a server capabilities request fails.
If this were encountered during a device reset for example,
SCRQ's may not be re-allocated. This operation is not necessary
anymore so remove it.
Signed-off-by: Thomas Falcon <tlfalcon@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: David S. Miller <davem@davemloft.net>

Tariq Toukan says:

====================
mlx4 misc fixes

This patchset contains misc bug fixes from the team
to the mlx4 Core and Eth drivers.

Patch 1 by Inbar fixes a wrong ethtool indication for Wake-on-LAN.
The other 3 patches by Jack add a missing capability description,
and fixes the off-by-1 misalignment for the following capabilities
descriptions.

Series generated against net commit:
cc75f851 samples/bpf: fix bpf tunnel cleanup
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

The cited commit introduced the following new enum value in file
include/linux/mlx4/device.h:

    QUERY_DEV_CAP_DIAG_RPRT_PER_PORT

However, it failed to introduce a corresponding entry in function
dump_dev_cap_flags2() for outputting a line in the message log
when this capability bit is set.

The change here fixes that omission.

Fixes: c7c122ed ("net/mlx4: Add diagnostic counters capability bit")
Reported-by: Mukesh Kacker <mukesh.kacker@oracle.com>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The cited commit introduced the following new enum value in file
include/linux/mlx4/device.h:

    MLX4_DEV_CAP_FLAG2_SVLAN_BY_QP

However the value of MLX4_DEV_CAP_FLAG2_SVLAN_BY_QP needs to stay
consistent with the value used in another namespace in
function dump_dev_cap_flags2(), which is manually kept in sync.
The change here restores that consistency.

Fixes: 7c3d21c8 ("net/mlx4_core: Preparation for VF vlan protocol 802.1ad")
Reported-by: Mukesh Kacker <mukesh.kacker@oracle.com>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The index value in function dump_dev_cap_flags2() for outputting
"sl to vl mapping table change event support" needs to be
consistent with the value of the enumerated constant
MLX4_DEV_CAP_FLAG2_SL_TO_VL_CHANGE_EVENT defined in file
include/linux/mlx4_device.h

The change here restores that consistency.

Fixes: fd10ed8e ("IB/mlx4: Fix possible vl/sl field mismatch in LRH header in QP1 packets")
Reported-by: Mukesh Kacker <mukesh.kacker@oracle.com>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently when WoL is supported but disabled, ethtool reports:
"Supports Wake-on: d".
Fix the indication of Wol support, so that the indication
remains "g" all the time if the NIC supports WoL.

Tested:
As accepted, when NIC supports WoL- ethtool reports:
	Supports Wake-on: g
	Wake-on: d
when NIC doesn't support WoL- ethtool reports:
        Supports Wake-on: d
        Wake-on: d

Fixes: 14c07b13 ("mlx4: Wake on LAN support")
Signed-off-by: Inbar Karmy <inbark@mellanox.com>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Nisar Sayed says:

====================
lan78xx: Fixes to lan78xx driver

This series of patches are for lan78xx driver.

These patches fixes potential issues associated with lan78xx driver
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Fix to handle hard_header_len update

When ifconfig up/down sequence is initiated hard_header_len
get updated incrementally for each ifconfig up /down sequence,
this leads invalid hard_header_len, moving to lan78xx_bind
to have one time update of hard_header_len addresses the issue.
Signed-off-by: Nisar Sayed <Nisar.Sayed@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

USB fast connect/disconnect crash fix

When USB plugged/unplugged at fast rate,
lan78xx_mdio_init() in lan78xx_bind() failing case is not handled.
Whenever  lan78xx_mdio_init() failed, dev->mdiobus will be freed, however
since lan78xx_bind() not consider as error and try to proceed for
further initialization in lan78xx_probe() which leads system hung/crash.
Also when register_netdev() failed, netdev is freed without calling lan78xx_unbind().
Hence halting the failed cases right manner fixes the system crash/hung issue.
Signed-off-by: Nisar Sayed <Nisar.Sayed@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Florian Fainelli says:

====================
drivers: net: Fix 64-bit statistics seqcount init

This patch series fixes a bunch of drivers to have their 64-bit statistics
seqcount cookie be initialized correctly. Most of these drivers (except b44,
gtp) are probably used on 64-bit only hosts and so the lockdep splat might have
never been seen.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

On 32-bit hosts and with CONFIG_DEBUG_LOCK_ALLOC we should be seeing a
lockdep splat indicating this seqcount is not correctly initialized, fix
that by using the proper helper function: netdev_alloc_pcpu_stats().

Fixes: 2ad7bf36 ("ipvlan: Initial check-in of the IPVLAN driver.")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

On 32-bit hosts and with CONFIG_DEBUG_LOCK_ALLOC we should be seeing a
lockdep splat indicating this seqcount is not correctly initialized, fix
that. In commit 6c80f3fc ("netvsc: report per-channel stats in
ethtool statistics") netdev_alloc_pcpu_stats() was removed in favor of
open-coding the 64-bits statistics, except that u64_stats_init() was
missed.

Fixes: 6c80f3fc ("netvsc: report per-channel stats in ethtool statistics")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

On 32-bit hosts and with CONFIG_DEBUG_LOCK_ALLOC we should be seeing a
lockdep splat indicating this seqcount is not correctly initialized, fix
that by using netdev_alloc_pcpu_stats() instead of an open coded
allocation.

Fixes: 459aa660 ("gtp: add initial driver for datapath of GPRS Tunneling Protocol (GTP-U)")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

On 32-bit hosts and with CONFIG_DEBUG_LOCK_ALLOC we should be seeing a
lockdep splat indicating this seqcount is not correctly initialized, fix
that.

Fixes: 4c352362 ("net: add driver for Netronome NFP4000/NFP6000 NIC VFs")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

On 32-bit hosts and with CONFIG_DEBUG_LOCK_ALLOC we should be seeing a
lockdep splat indicating this seqcount is not correctly initialized, fix
that.

Fixes: 4197aa7b ("ixgbevf: provide 64 bit statistics")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

On 32-bit hosts and with CONFIG_DEBUG_LOCK_ALLOC we should be seeing a
lockdep splat indicating this seqcount is not correctly initialized, fix
that.

Fixes: 980e9b11 ("i40e: Add support for 64 bit netstats")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>