For retransmitted packets, TCP needs to resort to using TCP timestamps
for computing RTT samples. In the common case where the data and ACK
fall in the same 1-millisecond interval, TCP senders with millisecond-
granularity TCP timestamps compute a ca_rtt_us of 0. This ca_rtt_us
of 0 propagates to rs->rtt_us.

This value of 0 can cause performance problems for congestion control
modules. For example, in BBR, the zero min_rtt sample can bring the
min_rtt and BDP estimate down to 0, reduce snd_cwnd and result in a
low throughput. It would be hard to mitigate this with filtering in
the congestion control module, because the proper floor to apply would
depend on the method of RTT sampling (using timestamp options or
internally-saved transmission timestamps).

This fix applies a floor of 1 for the RTT sample delta from TCP
timestamps, so that seq_rtt_us, ca_rtt_us, and rs->rtt_us will be at
least 1 * (USEC_PER_SEC / TCP_TS_HZ).

Note that the receiver RTT computation in tcp_rcv_rtt_measure() and
min_rtt computation in tcp_update_rtt_min() both already apply a floor
of 1 timestamp tick, so this commit makes the code more consistent in
avoiding this edge case of a value of 0.
Signed-off-by: NJianfeng Wang <jfwang@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Acked-by: NKevin Yang <yyd@google.com>
Acked-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Using is_broadcast_ether_addr() instead of directly use
memcmp() to determine if the ethernet address is broadcast
address.

spatch with a semantic match is used to found this problem.
(http://coccinelle.lip6.fr/)
Signed-off-by: NHuang Guobin <huangguobin4@huawei.com>
Acked-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Florinel Iordache says:

====================
DPAA FMan driver fixes

Here are several fixes for the DPAA FMan driver.

v2 changes:
* corrected patch 4 by removing the line added by mistake
* used longer fixes tags with the first 12 characters of the SHA-1 ID

v3 changes:
* remove the empty line inserted after fixes tag
====================
Acked-by: NMadalin Bucur <madalin.bucur@oss.nxp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix memory allocation for ethernet address hash table.
The code was wrongly allocating an array for eth hash table which
is incorrect because this is the main structure for eth hash table
(struct eth_hash_t) that contains inside a number of elements.

Fixes: 57ba4c9b ("fsl/fman: Add FMan MAC support")
Signed-off-by: NFlorinel Iordache <florinel.iordache@nxp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add a safe check to avoid dereferencing null pointer

Fixes: 57ba4c9b ("fsl/fman: Add FMan MAC support")
Signed-off-by: NFlorinel Iordache <florinel.iordache@nxp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The parameter 'priority' is incorrectly forced to zero which ultimately
induces logically dead code in the subsequent lines.

Fixes: 57ba4c9b ("fsl/fman: Add FMan MAC support")
Signed-off-by: NFlorinel Iordache <florinel.iordache@nxp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Check before using returned value to avoid dereferencing null pointer.

Fixes: 18a6c85f ("fsl/fman: Add FMan Port Support")
Signed-off-by: NFlorinel Iordache <florinel.iordache@nxp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Potentially overflowing expression (ts_freq << 16 and intgr << 16)
declared as type u32 (32-bit unsigned) is evaluated using 32-bit
arithmetic and then used in a context that expects an expression of
type u64 (64-bit unsigned) which ultimately is used as 16-bit
unsigned by typecasting to u16. Fixed by using an unsigned 32-bit
integer since the value is truncated anyway in the end.

Fixes: 414fd46e ("fsl/fman: Add FMan support")
Signed-off-by: NFlorinel Iordache <florinel.iordache@nxp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Pablo Neira Ayuso says:

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

1) UAF in chain binding support from previous batch, from Dan Carpenter.

2) Queue up delayed work to expire connections with no destination,
   from Andrew Sy Kim.

3) Use fallthrough pseudo-keyword, from Gustavo A. R. Silva.

4) Replace HTTP links with HTTPS, from Alexander A. Klimov.

5) Remove superfluous null header checks in ip6tables, from
   Gaurav Singh.

6) Add extended netlink error reporting for expression.

7) Report EEXIST on overlapping chain, set elements and flowtable
   devices.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Avoid a memset after a call to 'dma_alloc_coherent()'.
This is useless since
commit 518a2f19 ("dma-mapping: zero memory returned from dma_alloc_*")
Signed-off-by: NChristophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Update the size used in 'dma_free_coherent()' in order to match the one
used in the corresponding 'dma_alloc_coherent()', in
'spider_net_init_chain()'.

Fixes: d4ed8f8d ("Spidernet DMA coalescing")
Signed-off-by: NChristophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Update the size used in 'dma_free_coherent()' in order to match the one
used in the corresponding 'dma_alloc_coherent()'.

Fixes: 369a782a ("net: sgi: ioc3-eth: ensure tx ring is 16k aligned.")
Signed-off-by: NChristophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

On an error exit path, a negative error code should be returned
instead of a positive return value.

Fixes: 0c45d7fe ("liquidio: fix use of pf in pass-through mode in a virtual machine")
Cc: Rick Farrington <ricardo.farrington@cavium.com>
Signed-off-by: NTianjia Zhang <tianjia.zhang@linux.alibaba.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In the case of invalid rule, a positive value EINVAL is returned here.
I think this is a typo error. It is necessary to return an error value.

Cc: Po Liu <Po.Liu@nxp.com>
Signed-off-by: NTianjia Zhang <tianjia.zhang@linux.alibaba.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In function hw_atl_a0_hw_multicast_list_set(), when an invalid
request is encountered, a negative error code should be returned.

Fixes: bab6de8f ("net: ethernet: aquantia: Atlantic A0 and B0 specific functions")
Cc: David VomLehn <vomlehn@texas.net>
Signed-off-by: NTianjia Zhang <tianjia.zhang@linux.alibaba.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In queue_skb(), skb->data is mapped to streaming DMA on line 850:
  dma_map_single(..., skb->data, ...);

Then skb->data is accessed on lines 862 and 863:
  tbd->word_4 = (skb->data[0] << 24) | (skb->data[1] << 16) |
           (skb->data[2] <<  8) | (skb->data[3] <<  0);
and on lines 893 and 894:
  tbd->word_4 = (skb->data[0] << 24) | (skb->data[1] << 16) |
           (skb->data[2] <<  8) | (skb->data[3] <<  0);

These accesses may cause data inconsistency between CPU cache and
hardware.

To fix this problem, the calculation result of skb->data is stored in a
local variable before DMA mapping, and then the driver accesses this
local variable instead of skb->data.
Signed-off-by: NJia-Ju Bai <baijiaju@tsinghua.edu.cn>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In do_tx(), skb->data is mapped to streaming DMA on line 1111:
  paddr = dma_map_single(...,skb->data,DMA_TO_DEVICE);

Then skb->data is accessed on line 1153:
  (skb->data[3] & 0xf)

This access may cause data inconsistency between CPU cache and hardware.

To fix this problem, skb->data[3] is assigned to a local variable before
DMA mapping, and then the driver accesses this local variable instead of
skb->data[3].
Signed-off-by: NJia-Ju Bai <baijiaju@tsinghua.edu.cn>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

PHYLIB is not selected by the mvusb driver but it uses mdio devres
helpers. Explicitly select MDIO_DEVRES in this driver's Kconfig entry.
Reported-by: Nkernel test robot <lkp@intel.com>
Fixes: 1814cff2 ("net: phy: add a Kconfig option for mdio_devres")
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: NAndrew Lunn <andrew@lunn.ch>
Reviewed-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add a missing return statement to atalk_proc_init so it doesn't return
-ENOMEM when successful.  This allows the appletalk module to load
properly.

Fixes: e2bcd8b0 ("appletalk: use remove_proc_subtree to simplify procfs code")
Link: https://www.downtowndougbrown.com/2020/08/hacking-up-a-fix-for-the-broken-appletalk-kernel-module-in-linux-5-1-and-newer/Reported-by: NChristopher KOBAYASHI <chris@disavowed.jp>
Reported-by: NDoug Brown <doug@downtowndougbrown.com>
Signed-off-by: NVincent Duvert <vincent.ldev@duvert.net>
[lukas: add missing tags]
Signed-off-by: NLukas Wunner <lukas@wunner.de>
Cc: stable@vger.kernel.org # v5.1+
Cc: Yue Haibing <yuehaibing@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This adds full 802.1q VLAN support to the qca8k, allowing the use of
vlan_filtering and more complicated bridging setups than allowed by
basic port VLAN support.

Tested with a number of untagged ports with separate VLANs and then a
trunk port with all the VLANs tagged on it.

v3:
- Pull QCA8K_PORT_VID_DEF changes into separate cleanup patch
- Reverse Christmas tree notation for variable definitions
- Use untagged instead of tagged for consistency
v2:
- Return sensible errnos on failure rather than -1 (rmk)
- Style cleanups based on Florian's feedback
- Silently allow VLAN 0 as device correctly treats this as no tag
Signed-off-by: NJonathan McDowell <noodles@earth.li>
Reviewed-by: NFlorian Fainelli <f.fainelli@gmail.com>
Acked-by: NVladimir Oltean <olteanv@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Rather than using a magic value of 1 when configuring the port VIDs add
a QCA8K_PORT_VID_DEF define and use that instead. Also fix up the
bitmask in the process; the top 4 bits are reserved so this wasn't a
problem, but only masking 12 bits is the correct approach.
Signed-off-by: NJonathan McDowell <noodles@earth.li>
Reviewed-by: NFlorian Fainelli <f.fainelli@gmail.com>
Acked-by: NVladimir Oltean <olteanv@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Tony Nguyen says:

====================
100GbE Intel Wired LAN Driver Updates 2020-08-01

This series contains updates to the ice driver only.

Wei Yongjun marks power management functions with __maybe_unused.

Nick disables VLAN pruning in promiscuous mode and renames grst_delay to
grst_timeout.

Kiran modifies the check for linearization and corrects the vsi_id mask
value.

Vignesh replaces the use of flow profile locks to RSS profile locks for RSS
rule removal. Destroys flow profile lock on clearing XLT table and
clears extraction sequence entries.

Jesse adds some statistics and removes an unreported one.

Brett allows for 2 queue configuration for VFs.

Surabhi adds a check for failed allocation of an extraction sequence
table.

Tony updates the PTYPE lookup table and makes other trivial fixes.

Victor extends profile ID locks to be held until all references are
completed.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When we don't care about vlan depth, we could pass NULL instead of the
address of a unused local variable to skb_network_protocol() as a param.
Signed-off-by: NMiaohe Lin <linmiaohe@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In fact, skb_pagelen() - skb_headlen() is equal to __skb_pagelen(), use it
directly to avoid unnecessary skb_headlen() call.

Also fix the CHECK note of checkpatch.pl:
    Comparison to NULL could be written "!__pskb_pull_tail"
Signed-off-by: NMiaohe Lin <linmiaohe@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use eth_zero_addr() to clear mac address instead of memset().
Signed-off-by: NMiaohe Lin <linmiaohe@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use eth_zero_addr() to clear mac address instead of memset().
Signed-off-by: NMiaohe Lin <linmiaohe@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit c8729cac ("cxgb4: add ethtool n-tuple filter insertion")
has removed checking control key for determining IP address types
for TC-FLOWER rules, which causes all the rules being inserted to
hardware to become IPv6 rule type always. So, add back the check
to select the correct IP address type to extract and hence fix the
correct rule type being inserted to hardware.

Also, ethtool_rx_flow_key doesn't have any control key and instead
directly sets the IPv4/IPv6 address keys. So, explicitly set the
IP address type for ethtool n-tuple filters to reuse the same code.

Fixes: c8729cac ("cxgb4: add ethtool n-tuple filter insertion")
Signed-off-by: NRahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The flag indicating the selftest to run is a bitmask. So, fix the
check. Also, the selftests will fail if adapter initialization has
not been completed yet. So, add appropriate check and bail sooner.

Fixes: 7235ffae ("cxgb4: add loopback ethtool self-test")
Signed-off-by: NRahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Shannon Nelson says:

====================
ionic txrx updates

These are a few patches to do some cleanup in the packet
handling and give us more flexibility in tuning performance
by allowing us to put Tx handling on separate interrupts
when it makes sense for particular traffic loads.

v3: simplified queue count change logging, removed unnecessary
    check for no count change
v2: dropped the original patch 2 for ringsize change
    changed the separated tx/rx interrupts to use ethtool -L
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add the capability to split the Tx queues onto their own
interrupts with their own napi contexts.  This gives the
opportunity for more direct control of Tx interrupt
handling, such as CPU affinity and interrupt coalescing,
useful for some traffic loads.

v2: use ethtool -L, not a vendor specific priv-flag
v3: simplify logging, drop unnecessary "no-change" tests
Reviewed-by: NJakub Kicinski <kuba@kernel.org>
Signed-off-by: NShannon Nelson <snelson@pensando.io>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We give the tx clean path its own budget and service routine in
order to give a little more leeway to be more aggressive, and
in preparation for coming changes.  We've found this gives us
a little better performance in some packet processing scenarios
without hurting other scenarios.
Signed-off-by: NShannon Nelson <snelson@pensando.io>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We really don't need to hit the Rx queue doorbell so many times,
we can wait to the end and cause a little less thrash.
Signed-off-by: NShannon Nelson <snelson@pensando.io>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The GRE tunnel can be used to transport traffic that does not rely on a
Internet checksum (e.g. SCTP). The issue can be triggered creating a GRE
or GRETAP tunnel and transmitting SCTP traffic ontop of it where CRC
offload has been disabled. In order to fix the issue we need to
recompute the GRE csum in gre_gso_segment() not relying on the inner
checksum.
The issue is still present when we have the CRC offload enabled.
In this case we need to disable the CRC offload if we require GRE
checksum since otherwise skb_checksum() will report a wrong value.

Fixes: 90017acc ("sctp: Add GSO support")
Signed-off-by: NLorenzo Bianconi <lorenzo@kernel.org>
Reviewed-by: NMarcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We need to clear all of the bridge private skb variables as they can be
stale due to the packet being recirculated through the stack and then
transmitted through the bridge device. Similar memset is already done on
bridge's input. We've seen cases where proxyarp_replied was 1 on routed
multicast packets transmitted through the bridge to ports with neigh
suppress which were getting dropped. Same thing can in theory happen with
the port isolation bit as well.

Fixes: 821f1b21 ("bridge: add new BR_NEIGH_SUPPRESS port flag to suppress arp and nd flood")
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

"how" was used as a boolean. Change the type to bool, and improve
variable name
Signed-off-by: NFlorent Fourcot <florent.fourcot@wifirst.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Second parameter of addrconf_ifdown "how" is used as a boolean
internally. It does not make sense to call it with something different
of 0 or 1.

This value is set to 2 in all git history.
Signed-off-by: NFlorent Fourcot <florent.fourcot@wifirst.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Eelco Chaudron says:

====================
net: openvswitch: masks cache enhancements

This patchset adds two enhancements to the Open vSwitch masks cache.

Changes in v4 [patch 2/2 only]:
 - Remove null check before calling free_percpu()
 - Make ovs_dp_change() return appropriate error codes

Changes in v3 [patch 2/2 only]:
 - Use is_power_of_2() function
 - Use array_size() function
 - Fix remaining sparse errors

Changes in v2 [patch 2/2 only]:
 - Fix sparse warnings
 - Fix netlink policy items reported by Florian Westphal
====================
Signed-off-by: NEelco Chaudron <echaudro@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch makes the masks cache size configurable, or with
a size of 0, disable it.
Reviewed-by: NPaolo Abeni <pabeni@redhat.com>
Reviewed-by: NTonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: NEelco Chaudron <echaudro@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add a counter that counts the number of masks cache hits, and
export it through the megaflow netlink statistics.
Reviewed-by: NPaolo Abeni <pabeni@redhat.com>
Reviewed-by: NTonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: NEelco Chaudron <echaudro@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Release skb memory in mvpp2_rx() if mvpp2_rx_refill routine fails

Fixes: b5015854 ("net: mvpp2: fix refilling BM pools in RX path")
Signed-off-by: NLorenzo Bianconi <lorenzo@kernel.org>
Acked-by: NMatteo Croce <mcroce@microsoft.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>