OSDN Git Service

uclinux-h8/linux.git
6 years agovxlan: fix default fdb entry netlink notify ordering during netdev create
Roopa Prabhu [Wed, 4 Jul 2018 23:46:32 +0000 (16:46 -0700)]
vxlan: fix default fdb entry netlink notify ordering during netdev create

Problem:
In vxlan_newlink, a default fdb entry is added before register_netdev.
The default fdb creation function also notifies user-space of the
fdb entry on the vxlan device which user-space does not know about yet.
(RTM_NEWNEIGH goes before RTM_NEWLINK for the same ifindex).

This patch fixes the user-space netlink notification ordering issue
with the following changes:
- decouple fdb notify from fdb create.
- Move fdb notify after register_netdev.
- Call rtnl_configure_link in vxlan newlink handler to notify
userspace about the newlink before fdb notify and
hence avoiding the user-space race.

Fixes: afbd8bae9c79 ("vxlan: add implicit fdb entry for default destination")
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agovxlan: make netlink notify in vxlan_fdb_destroy optional
Roopa Prabhu [Wed, 4 Jul 2018 23:46:31 +0000 (16:46 -0700)]
vxlan: make netlink notify in vxlan_fdb_destroy optional

Add a new option do_notify to vxlan_fdb_destroy to make
sending netlink notify optional. Used by a later patch.

Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agovxlan: add new fdb alloc and create helpers
Roopa Prabhu [Wed, 4 Jul 2018 23:46:30 +0000 (16:46 -0700)]
vxlan: add new fdb alloc and create helpers

- Add new vxlan_fdb_alloc helper
- rename existing vxlan_fdb_create into vxlan_fdb_update:
        because it really creates or updates an existing
        fdb entry
- move new fdb creation into a separate vxlan_fdb_create

Main motivation for this change is to introduce the ability
to decouple vxlan fdb creation and notify, used in a later patch.

Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agortnetlink: add rtnl_link_state check in rtnl_configure_link
Roopa Prabhu [Wed, 4 Jul 2018 23:46:29 +0000 (16:46 -0700)]
rtnetlink: add rtnl_link_state check in rtnl_configure_link

rtnl_configure_link sets dev->rtnl_link_state to
RTNL_LINK_INITIALIZED and unconditionally calls
__dev_notify_flags to notify user-space of dev flags.

current call sequence for rtnl_configure_link
rtnetlink_newlink
    rtnl_link_ops->newlink
    rtnl_configure_link (unconditionally notifies userspace of
                         default and new dev flags)

If a newlink handler wants to call rtnl_configure_link
early, we will end up with duplicate notifications to
user-space.

This patch fixes rtnl_configure_link to check rtnl_link_state
and call __dev_notify_flags with gchanges = 0 if already
RTNL_LINK_INITIALIZED.

Later in the series, this patch will help the following sequence
where a driver implementing newlink can call rtnl_configure_link
to initialize the link early.

makes the following call sequence work:
rtnetlink_newlink
    rtnl_link_ops->newlink (vxlan) -> rtnl_configure_link (initializes
                                                link and notifies
                                                user-space of default
                                                dev flags)
    rtnl_configure_link (updates dev flags if requested by user ifm
                         and notifies user-space of new dev flags)

Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodevlink: fix incorrect return statement
Arnd Bergmann [Fri, 6 Jul 2018 12:58:51 +0000 (14:58 +0200)]
devlink: fix incorrect return statement

A newly added dummy helper function tries to return a failure from a "void"
function:

In file included from include/net/dsa.h:24,
                 from arch/arm/plat-orion/common.c:21:
include/net/devlink.h: In function 'devlink_param_value_changed':
include/net/devlink.h:771:9: error: 'return' with a value, in function returning void [-Werror]
  return -EOPNOTSUPP;

This fixes it by removing the bogus statement.

Fixes: ea601e170988 ("devlink: Add devlink notifications support for params")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlxsw: spectrum_router: avoid uninitialized variable access
Arnd Bergmann [Fri, 6 Jul 2018 12:44:45 +0000 (14:44 +0200)]
mlxsw: spectrum_router: avoid uninitialized variable access

When CONFIG_BRIDGE_VLAN_FILTERING is disabled, gcc correctly points out
that the 'vid' variable is uninitialized whenever br_vlan_get_pvid
returns an error:

drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c: In function 'mlxsw_sp_rif_vlan_fid_get':
drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c:6881:6: error: 'vid' may be used uninitialized in this function [-Werror=maybe-uninitialized]

This changes the condition check to always return -EINVAL here,
which I guess is what the author intended here.

Fixes: e6f1960ae6c7 ("mlxsw: spectrum_router: Allocate FID according to PVID")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'hns3-misc-fixes'
David S. Miller [Sat, 7 Jul 2018 02:13:07 +0000 (11:13 +0900)]
Merge branch 'hns3-misc-fixes'

Salil Mehta says:

====================
Misc. bug fixes & cleanups for HNS3 driver

This patch-set presents some miscellaneous bug fixes and cleanups
for the HNS3 driver.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Prevent sending command during global or core reset
Huazhong Tan [Fri, 6 Jul 2018 10:28:04 +0000 (11:28 +0100)]
net: hns3: Prevent sending command during global or core reset

According to hardware's description, driver should not send command to
IMP while hardware doing global or core reset.

Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Remove the warning when clear reset cause
Peng Li [Fri, 6 Jul 2018 10:28:03 +0000 (11:28 +0100)]
net: hns3: Remove the warning when clear reset cause

Only the core/global/IMP reset need clear cause, other type does not
need do it. The warning may be treated as error as it is normal. This
patch removes the warning.

Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Fix get_vector ops in hclgevf_main module
Yunsheng Lin [Fri, 6 Jul 2018 10:28:02 +0000 (11:28 +0100)]
net: hns3: Fix get_vector ops in hclgevf_main module

The hclgevf_free_vector function expects the caller to pass
the vector_id to it, and hclgevf_put_vector pass vector to
it now, which will cause vector allocation problem.

This patch fixes it by converting vector into vector_id before
calling hclgevf_free_vector.

Fixes: e2cb1dec9779 ("net: hns3: Add HNS3 VF HCL(Hardware Compatibility Layer) Support")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Fix warning bug when doing lp selftest
Yunsheng Lin [Fri, 6 Jul 2018 10:28:01 +0000 (11:28 +0100)]
net: hns3: Fix warning bug when doing lp selftest

The napi_alloc_skb is excepted to be called under the
non-preemptible code path when it is called by hns3_clean_rx_ring
during loopback selftest, otherwise the below warning will be
logged:

[   92.420780] BUG: using smp_processor_id() in preemptible
[00000000] code: ethtool/1873
<SNIP>
[   92.463202]  check_preemption_disabled+0xf8/0x100
[   92.467893]  debug_smp_processor_id+0x1c/0x28
[   92.472239]  __napi_alloc_skb+0x30/0x130
[   92.476158]  hns3_clean_rx_ring+0x118/0x5f0 [hns3]
[   92.480941]  hns3_self_test+0x32c/0x4d0 [hns3]
[   92.485375]  ethtool_self_test+0xdc/0x1e8
[   92.489372]  dev_ethtool+0x1020/0x1da8
[   92.493109]  dev_ioctl+0x188/0x3a0
[   92.496499]  sock_do_ioctl+0xf4/0x208
[   92.500148]  sock_ioctl+0x228/0x3e8
[   92.503626]  do_vfs_ioctl+0xc4/0x880
[   92.507189]  SyS_ioctl+0x94/0xa8
[   92.510404]  el0_svc_naked+0x30/0x34

This patch fix it by disabling preemption when calling
hns3_clean_rx_ring during loopback selftest.

Fixes: c39c4d98dc65 ("net: hns3: Add mac loopback selftest support in hns3 driver")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Add configure for mac minimal frame size
Jian Shen [Fri, 6 Jul 2018 10:28:00 +0000 (11:28 +0100)]
net: hns3: Add configure for mac minimal frame size

When change the mtu, the minimal frame size of mac will be set
to zero, it is incorrect. This patch fixes it by set it to the
default value.

Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Fix for mailbox message truncated problem
Fuyun Liang [Fri, 6 Jul 2018 10:27:59 +0000 (11:27 +0100)]
net: hns3: Fix for mailbox message truncated problem

The payload of mailbox message is 16 byte and the value of
HCLGE_MBX_MAX_ARQ_MSG_SIZE is 8. A message truncated problem will
happen when mailbox message is converted to ARQ message. This patch
replaces HCLGE_MBX_MAX_ARQ_MSG_SIZE with the size of ARQ message in
hclgevf_mbx_handler to fix this problem.

Fixes: b11a0bb231f3 ("net: hns3: Add mailbox support to VF driver")
Signed-off-by: Fuyun Liang <liangfuyun1@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Fix for l4 checksum offload bug
Yunsheng Lin [Fri, 6 Jul 2018 10:27:58 +0000 (11:27 +0100)]
net: hns3: Fix for l4 checksum offload bug

Hardware only support tcp/udp/sctp l4 checksum offload, but
the driver currently tell hardware to do l4 checksum offlad when
l3 is IPv4 or IPv6, which may cause checksumm error.

This patch fixes it by only enabling the l4 offload when l4 is
tcp/udp/sctp.

Fixes: 76ad4f0ee747 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Fix for waterline not setting correctly
Yunsheng Lin [Fri, 6 Jul 2018 10:27:57 +0000 (11:27 +0100)]
net: hns3: Fix for waterline not setting correctly

The HCLGE_RX_PRIV_EN_B is used to tell the firmware whether
to update the specific waterline value, if the is not set,
the firmware will ignore the value.

This patch fixes by setting the HCLGE_RX_PRIV_EN_B even if
the updated value is zero.

Fixes: 46a3df9f9718 ("net: hns3: Add HNS3 Acceleration Engine & Compatibility Layer Support")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Fix for mac pause not disable in pfc mode
Yunsheng Lin [Fri, 6 Jul 2018 10:27:56 +0000 (11:27 +0100)]
net: hns3: Fix for mac pause not disable in pfc mode

When pfc pause mode is enable, the mac pause mode need to be
disabled, otherwise the pfc pause packet will not be sent when
congestion happens.

This patch fixes by disabling the mac pause when pfc pause is
enabled.

Fixes: 848440544b41 ("net: hns3: Add support of TX Scheduler & Shaper to HNS3 driver")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Fix tc setup when netdev is first up
Yunsheng Lin [Fri, 6 Jul 2018 10:27:55 +0000 (11:27 +0100)]
net: hns3: Fix tc setup when netdev is first up

Currently, tc related configuration is not setup when the
netdev is first up, which cause the stack only using tc 0
problem.

This patch fixes it by setting the tc related configuration
using the info from NCL_CONFIG when netdev is first up.

Fixes: 76ad4f0ee747 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoliquidio: fix kernel panic when NIC firmware is older than 1.7.2
Rick Farrington [Fri, 6 Jul 2018 18:27:07 +0000 (11:27 -0700)]
liquidio: fix kernel panic when NIC firmware is older than 1.7.2

Pre-1.7.2 NIC firmware does not support (and does not respond to) the "get
speed" command which is sent by the 1.7.2 driver (for CN23XX-225 cards
only) during modprobe.  Due to a bug in older firmware (with respect to
unknown commands), this unsupported command causes a cascade of errors that
ends in a kernel panic.

Fix it by making the sending of the "get speed" command conditional on the
firmware version.

Signed-off-by: Rick Farrington <ricardo.farrington@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'sock-cookie-initializers'
David S. Miller [Sat, 7 Jul 2018 01:58:50 +0000 (10:58 +0900)]
Merge branch 'sock-cookie-initializers'

Willem de Bruijn says:

====================
sock cookie initializers

Recent UDP GSO and SO_TXTIME features added new fields to cookie
structs.

When adding a field, all sites where a struct is initialized have to
be updated, which is a lot of boilerplate. Alternatively, a field can
be initialized selectively, but this is fragile. I introduced a bug
in udp gso where an uninitialized field was read. See also fix commit
("9887cba19978 ip: limit use of gso_size to udp").

Introduce initializers for structs ipcm(6)_cookie and sockc_cookie.

patch 1..3 do exactly this.
patch 4..5 make ipv4 and ipv6 handle cookies the same way and
           remove some boilerplate in doing so.
patch 6    removes the udp gso branch that needed the above fix
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoip: unconditionally set cork gso_size
Willem de Bruijn [Fri, 6 Jul 2018 14:12:59 +0000 (10:12 -0400)]
ip: unconditionally set cork gso_size

Now that ipc(6)->gso_size is correctly initialized in all callers of
ip(6)_setup_cork, it is safe to unconditionally pass it to the cork.

Link: http://lkml.kernel.org/r/20180619164752.143249-1-willemdebruijn.kernel@gmail.com
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoip: remove tx_flags from ipcm_cookie and use same logic for v4 and v6
Willem de Bruijn [Fri, 6 Jul 2018 14:12:58 +0000 (10:12 -0400)]
ip: remove tx_flags from ipcm_cookie and use same logic for v4 and v6

skb_shinfo(skb)->tx_flags is derived from sk->sk_tsflags, possibly
after modification by __sock_cmsg_send, by calling sock_tx_timestamp.

The IPv4 and IPv6 paths do this conversion differently. In IPv4, the
individual protocols that support tx timestamps call this function
and store the result in ipc.tx_flags. In IPv6, sock_tx_timestamp is
called in __ip6_append_data.

There is no need to store both tx_flags and ts_flags in the cookie
as one is derived from the other. Convert when setting up the cork
and remove the redundant field. This is similar to IPv6, only have
the conversion happen only once per datagram, in ip(6)_setup_cork.

Also change __ip6_append_data to match __ip_append_data. Only update
tskey if timestamping is enabled with OPT_ID. The SOCK_.. test is
redundant: only valid protocols can have non-zero cork->tx_flags.

After this change the IPv4 and IPv6 logic is the same.

Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoipv6: fold sockcm_cookie into ipcm6_cookie
Willem de Bruijn [Fri, 6 Jul 2018 14:12:57 +0000 (10:12 -0400)]
ipv6: fold sockcm_cookie into ipcm6_cookie

ipcm_cookie includes sockcm_cookie. Do the same for ipcm6_cookie.

This reduces the number of arguments that need to be passed around,
applies ipcm6_init to all cookie fields at once and reduces code
differentiation between ipv4 and ipv6.

Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agosock: sockc cookie initializer
Willem de Bruijn [Fri, 6 Jul 2018 14:12:56 +0000 (10:12 -0400)]
sock: sockc cookie initializer

Initialize the cookie in one location to reduce code duplication and
avoid bugs from inconsistent initialization, such as that fixed in
commit 9887cba19978 ("ip: limit use of gso_size to udp").

Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoipv6: ipcm6_cookie initializer
Willem de Bruijn [Fri, 6 Jul 2018 14:12:55 +0000 (10:12 -0400)]
ipv6: ipcm6_cookie initializer

Initialize the cookie in one location to reduce code duplication and
avoid bugs from inconsistent initialization, such as that fixed in
commit 9887cba19978 ("ip: limit use of gso_size to udp").

Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoipv4: ipcm_cookie initializers
Willem de Bruijn [Fri, 6 Jul 2018 14:12:54 +0000 (10:12 -0400)]
ipv4: ipcm_cookie initializers

Initialize the cookie in one location to reduce code duplication and
avoid bugs from inconsistent initialization, such as that fixed in
commit 9887cba19978 ("ip: limit use of gso_size to udp").

Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'IP-listification-follow-ups'
David S. Miller [Fri, 6 Jul 2018 02:19:08 +0000 (11:19 +0900)]
Merge branch 'IP-listification-follow-ups'

Edward Cree says:

====================
IP listification follow-ups

While working on IPv6 list processing, I found another bug in the IPv4
 version.  So this patch series has that fix, and the IPv6 version with
 both fixes incorporated.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ipv6: listify ipv6_rcv() and ip6_rcv_finish()
Edward Cree [Thu, 5 Jul 2018 14:49:42 +0000 (15:49 +0100)]
net: ipv6: listify ipv6_rcv() and ip6_rcv_finish()

Essentially the same as the ipv4 equivalents.

Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ipv4: fix list processing on L3 slave devices
Edward Cree [Thu, 5 Jul 2018 14:47:39 +0000 (15:47 +0100)]
net: ipv4: fix list processing on L3 slave devices

If we have an L3 master device, l3mdev_ip_rcv() will steal the skb, but
 we were returning NET_RX_SUCCESS from ip_rcv_finish_core() which meant
 that ip_list_rcv_finish() would keep it on the list.  Instead let's
 move the l3mdev_ip_rcv() call into the caller, so that our response to
 a steal can be different in the single packet path (return
 NET_RX_SUCCESS) and the list path (forget this packet and continue).

Fixes: 5fa12739a53d ("net: ipv4: listify ip_rcv_finish")
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: core: filter: mark expected switch fall-through
Gustavo A. R. Silva [Wed, 4 Jul 2018 22:34:37 +0000 (17:34 -0500)]
net: core: filter: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Warning level 2 was used: -Wimplicit-fallthrough=2

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: decnet: dn_nsp_in: mark expected switch fall-through
Gustavo A. R. Silva [Wed, 4 Jul 2018 22:05:00 +0000 (17:05 -0500)]
net: decnet: dn_nsp_in: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agotipc: mark expected switch fall-throughs
Gustavo A. R. Silva [Wed, 4 Jul 2018 21:13:59 +0000 (16:13 -0500)]
tipc: mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Warning level 2 was used: -Wimplicit-fallthrough=2

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agocxgb4: Fix the condition to check if the card is T5
Ganesh Goudar [Wed, 4 Jul 2018 12:19:33 +0000 (17:49 +0530)]
cxgb4: Fix the condition to check if the card is T5

Use 'chip_ver' rather than 'chip' to check if the card
is T5.

Fixes: e8d452923ae6 ("cxgb4: clean up init_one")
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ethernet: sun: remove redundant variables adv and lpa and mii_reads
Colin Ian King [Thu, 5 Jul 2018 11:05:25 +0000 (12:05 +0100)]
net: ethernet: sun: remove redundant variables adv and lpa and mii_reads

Variables adv and lpa are being assigned but are never used hence they
are redundant and can be removed.  Also remove the unncessary mii_reads
too.

Cleans up clang warnings:
warning: variable 'lpa' set but not used [-Wunused-but-set-variable]
warning: variable 'adv' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'net-ethernet-Miscellaneous-Kconfig-and-Makefile-cleanups'
David S. Miller [Thu, 5 Jul 2018 11:05:55 +0000 (20:05 +0900)]
Merge branch 'net-ethernet-Miscellaneous-Kconfig-and-Makefile-cleanups'

Geert Uytterhoeven says:

====================
net: ethernet: Miscellaneous Kconfig and Makefile cleanups

This patch series contains a few Kconfig and Makefile cleanups w.r.t.
Ethernet vendors.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ethernet: sfc: Make subdir logic consistent with other vendors
Geert Uytterhoeven [Wed, 4 Jul 2018 11:50:13 +0000 (13:50 +0200)]
net: ethernet: sfc: Make subdir logic consistent with other vendors

Both SFC and SFC_FALCON depend on NET_VENDOR_SOLARFLARE, hence use the
latter to decide whether to descend into the sfc subdirectory.
Move the rule to descend into sfc/falcon to the sfc subdirectory.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Martin Habets <mhabets@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ethernet: Add missing VENDOR to Cadence and Packet Engines symbols
Geert Uytterhoeven [Wed, 4 Jul 2018 11:50:12 +0000 (13:50 +0200)]
net: ethernet: Add missing VENDOR to Cadence and Packet Engines symbols

The vendor guard Kconfig symbols for Cadence and Packet Engines use a
"NET_" prefix, while all other vendor guards use a "NET_VENDOR_"
prefix.  Hence make them consistent with the rest, and add the missing
trailing "S" for Packet Engines while at it.

As these options don't directly affect the kernel build, and default to
"y", this change has no impact on kernels built with existing
(def)configs.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ethernet: Make NET_VENDOR_AURORA default to yes
Geert Uytterhoeven [Wed, 4 Jul 2018 11:50:11 +0000 (13:50 +0200)]
net: ethernet: Make NET_VENDOR_AURORA default to yes

Enabling NET_VENDOR_* Kconfig options does not directly affect the
kernel build.  Hence NET_VENDOR_AURORA should default to yes, like other
NET_VENDOR_* options.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqlogic: netxen: remove various redundant variables
Colin Ian King [Wed, 4 Jul 2018 11:45:53 +0000 (12:45 +0100)]
qlogic: netxen: remove various redundant variables

Variables consumer, cmd_desc, end_cnt and no_of_desc are being assigned
but are never used hence they are redundant and can be removed.

Cleans up clang warnings:
warning: variable 'consumer' set but not used [-Wunused-but-set-variable]
warning: variable 'cmd_desc' set but not used [-Wunused-but-set-variable]
warning: variable 'end_cnt' set but not used [-Wunused-but-set-variable]
warning: variable 'no_of_desc' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'devlink-Add-configuration-parameters-support'
David S. Miller [Thu, 5 Jul 2018 10:58:36 +0000 (19:58 +0900)]
Merge branch 'devlink-Add-configuration-parameters-support'

Moshe Shemesh says:

====================
Add configuration parameters support

Add configuration parameters setting through devlink.
Each device registers supported configuration parameters table.
Each parameter can be either generic or driver specific.
The user can retrieve data on these parameters by "devlink param show"
command and can set new value to a parameter by "devlink param set"
command.
The parameters can be set in different configuration modes:
  runtime - set while driver is running, no reset required.
  driverinit - applied while driver initializes, requires restart
               driver by devlink reload command.
  permanent - written to device's non-volatile memory, hard reset required.

The patches at the end of the patchset introduce few params that are using
the introduced infrastructure on mlx4 and bnxt.

Command examples and output:

pci/0000:81:00.0:
  name internal_error_reset type generic
    values:
      cmode runtime value true
      cmode driverinit value true
  name max_macs type generic
    values:
      cmode driverinit value 128
  name enable_64b_cqe_eqe type driver-specific
    values:
      cmode driverinit value true
  name enable_4k_uar type driver-specific
    values:
      cmode driverinit value false

pci/0000:81:00.0:
  name internal_error_reset type generic
    values:
      cmode runtime value false
      cmode driverinit value true
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agobnxt_en: Add bnxt_en initial params table and register it.
Vasundhara Volam [Wed, 4 Jul 2018 11:30:37 +0000 (14:30 +0300)]
bnxt_en: Add bnxt_en initial params table and register it.

Create initial devlink parameters table for bnxt_en.
Table consists of a permanent generic parameter.

enable_sriov - Enables Single-Root Input/Output Virtualization(SR-IOV)
characteristic of the device.

Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodevlink: Add enable_sriov boolean generic parameter
Vasundhara Volam [Wed, 4 Jul 2018 11:30:36 +0000 (14:30 +0300)]
devlink: Add enable_sriov boolean generic parameter

enable_sriov - Enables Single-Root Input/Output Virtualization(SR-IOV)
characteristic of the device.

Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlx4: Add support for devlink reload and load driverinit values
Moshe Shemesh [Wed, 4 Jul 2018 11:30:35 +0000 (14:30 +0300)]
mlx4: Add support for devlink reload and load driverinit values

Add mlx4_devlink_reload() to support devlink reload operation.
Add mlx4_devlink_param_load_driverinit_values() to load values which
were set using driverinit configuration mode.

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlx4: Add mlx4 initial parameters table and register it
Moshe Shemesh [Wed, 4 Jul 2018 11:30:34 +0000 (14:30 +0300)]
mlx4: Add mlx4 initial parameters table and register it

Create initial parameters table for mlx4.
The table consists of two generic parameters and two driver-specific
parameters.
Generic:
  internal_err_reset - Enable reset device on internal errors. This
  parameter can be configured on mlx4 either on runtime or during driver
  initialization.
  max_macs - Max number of MACs per ETH port. For mlx4 this parameter
  value range is between 1 and 128. This parameter can be configured on
  mlx4 only during driver initialization.
Driver specific:
  enable_64b_cqe_eqe - Enable 64 byte CQEs/EQEs when the FW supports it.
  This parameter can be configured on mlx4 only during driver
  initialization.
  enable_4k_uar - Enable using 4K UAR. This parameter can be configured on
  mlx4 only during driver initialization.

Register the parameters table on mlx4_init_one() and unregister on
mlx4_remove_one().

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodevlink: Add generic parameters internal_err_reset and max_macs
Moshe Shemesh [Wed, 4 Jul 2018 11:30:33 +0000 (14:30 +0300)]
devlink: Add generic parameters internal_err_reset and max_macs

Add 2 first generic parameters to devlink configuration parameters set:
internal_err_reset - When set enables reset device on internal errors.
max_macs - max number of MACs per ETH port.

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodevlink: Add devlink notifications support for params
Moshe Shemesh [Wed, 4 Jul 2018 11:30:32 +0000 (14:30 +0300)]
devlink: Add devlink notifications support for params

Add devlink_param_notify() function to support devlink param notifications.
Add notification call to devlink param set, register and unregister
functions.
Add devlink_param_value_changed() function to enable the driver notify
devlink on value change. Driver should use this function after value was
changed on any configuration mode part to driverinit.

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodevlink: Add support for get/set driverinit value
Moshe Shemesh [Wed, 4 Jul 2018 11:30:31 +0000 (14:30 +0300)]
devlink: Add support for get/set driverinit value

"driverinit" configuration mode value is held by devlink to enable
the driver query the value after reload. Two additional functions
added to help the driver get/set the value from/to devlink:
devlink_param_driverinit_value_set() and
devlink_param_driverinit_value_get().

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodevlink: Add param set command
Moshe Shemesh [Wed, 4 Jul 2018 11:30:30 +0000 (14:30 +0300)]
devlink: Add param set command

Add param set command to set value for a parameter.
Value can be set to any of the supported configuration modes.

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodevlink: Add param get command
Moshe Shemesh [Wed, 4 Jul 2018 11:30:29 +0000 (14:30 +0300)]
devlink: Add param get command

Add param get command which gets data per parameter.
Option to dump the parameters data per device.

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodevlink: Add devlink_param register and unregister
Moshe Shemesh [Wed, 4 Jul 2018 11:30:28 +0000 (14:30 +0300)]
devlink: Add devlink_param register and unregister

Define configuration parameters data structure.
Add functions to register and unregister the driver supported
configuration parameters table.
For each parameter registered, the driver should fill all the parameter's
fields. In case the only supported configuration mode is "driverinit"
the parameter's get()/set() functions are not required and should be set
to NULL, for any other configuration mode, these functions are required
and should be set by the driver.

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/hamradio/6pack: remove redundant variable channel
Colin Ian King [Thu, 5 Jul 2018 10:11:07 +0000 (11:11 +0100)]
net/hamradio/6pack: remove redundant variable channel

Variable channel is being assigned but is never used hence it is
redundant and can be removed.

Cleans up two clang warnings:
warning: variable 'channel' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agofjes: use currently unused variable my_epid and max_epid
Colin Ian King [Thu, 5 Jul 2018 10:05:32 +0000 (11:05 +0100)]
fjes: use currently unused variable my_epid and max_epid

Variables my_epid and max_epid are currently assigned and not being
used - however, I suspect they were intended to be used in the for-loops
to reduce the dereferencing of hw.  Replace hw->my_epid and hw->max_epid
with these variables.

Cleans up clang warnings:
warning: variable 'my_epid' set but not used [-Wunused-but-set-variable]
variable 'max_epid' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: tehuti: remove redundant pointer skb
Colin Ian King [Thu, 5 Jul 2018 09:55:49 +0000 (10:55 +0100)]
net: tehuti: remove redundant pointer skb

Pointer skb is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'skb' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: socionext: remove redundant pointer ndev
Colin Ian King [Thu, 5 Jul 2018 09:26:13 +0000 (10:26 +0100)]
net: socionext: remove redundant pointer ndev

Pointer ndev is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'ndev' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: aquantia: Make some functions static
Wei Yongjun [Thu, 5 Jul 2018 09:00:10 +0000 (09:00 +0000)]
net: aquantia: Make some functions static

Fixes the following sparse warnings:

drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c:525:5: warning:
 symbol 'hw_atl_utils_mpi_set_speed' was not declared. Should it be static?
drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c:536:5: warning:
 symbol 'hw_atl_utils_mpi_set_state' was not declared. Should it be static?

Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: vsc73xx: Make some functions static
Wei Yongjun [Thu, 5 Jul 2018 08:59:09 +0000 (08:59 +0000)]
net: dsa: vsc73xx: Make some functions static

Fixes the following sparse warnings:

drivers/net/dsa/vitesse-vsc73xx.c:1054:6: warning:
 symbol 'vsc73xx_get_strings' was not declared. Should it be static?
drivers/net/dsa/vitesse-vsc73xx.c:1113:5: warning:
 symbol 'vsc73xx_get_sset_count' was not declared. Should it be static?
drivers/net/dsa/vitesse-vsc73xx.c:1122:6: warning:
 symbol 'vsc73xx_get_ethtool_stats' was not declared. Should it be static?

Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: fix spelling mistake "waitting" -> "waiting"
Colin Ian King [Thu, 5 Jul 2018 08:30:04 +0000 (09:30 +0100)]
net: dsa: fix spelling mistake "waitting" -> "waiting"

Trivial fix to spelling mistake in dev_err error message.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: limit each hash list length to MAX_GRO_SKBS
Li RongQing [Thu, 5 Jul 2018 06:34:32 +0000 (14:34 +0800)]
net: limit each hash list length to MAX_GRO_SKBS

After commit 07d78363dcff ("net: Convert NAPI gro list into a small hash
table.")' there is 8 hash buckets, which allows more flows to be held for
merging.  but MAX_GRO_SKBS, the total held skb for merging, is 8 skb still,
limit the hash table performance.

keep MAX_GRO_SKBS as 8 skb, but limit each hash list length to 8 skb, not
the total 8 skb

Signed-off-by: Li RongQing <lirongqing@baidu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agor8169: fix runtime suspend
Heiner Kallweit [Wed, 4 Jul 2018 19:11:29 +0000 (21:11 +0200)]
r8169: fix runtime suspend

When runtime-suspending we configure WoL w/o touching saved_wolopts.
If saved_wolopts == 0 we would power down the PHY in this case what's
wrong. Therefore we have to check the actual chip WoL settings here.

Fixes: 433f9d0ddcc6 ("r8169: improve saved_wolopts handling")
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ipv4: fix drop handling in ip_list_rcv() and ip_list_rcv_finish()
Edward Cree [Wed, 4 Jul 2018 18:23:50 +0000 (19:23 +0100)]
net: ipv4: fix drop handling in ip_list_rcv() and ip_list_rcv_finish()

Since callees (ip_rcv_core() and ip_rcv_finish_core()) might free or steal
 the skb, we can't use the list_cut_before() method; we can't even do a
 list_del(&skb->list) in the drop case, because skb might have already been
 freed and reused.
So instead, take each skb off the source list before processing, and add it
 to the sublist afterwards if it wasn't freed or stolen.

Fixes: 5fa12739a53d net: ipv4: listify ip_rcv_finish
Fixes: 17266ee93984 net: ipv4: listified version of ip_rcv
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agocxgb4: Add support to read actual provisioned resources
Casey Leedom [Wed, 4 Jul 2018 09:42:56 +0000 (15:12 +0530)]
cxgb4: Add support to read actual provisioned resources

In highly constrained resources environments (like the 124VF
T5 and 248VF T6 configurations), PF4 may not have very many
resources at all and we need to adapt to whatever we've been
allocated, this patch adds support to get the provisioned
resources.

Signed-off-by: Casey Leedom <leedom@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoepic100: remove redundant variable 'irq'
Colin Ian King [Wed, 4 Jul 2018 12:19:29 +0000 (13:19 +0100)]
epic100: remove redundant variable 'irq'

Variable 'irq' is being assigned but is never used hence it is
and can be removed.

Cleans up clang warning:
warning: variable 'irq' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agosfc: remove redundant variable old_vlan
Colin Ian King [Wed, 4 Jul 2018 12:13:01 +0000 (13:13 +0100)]
sfc: remove redundant variable old_vlan

Variable old_vlan is being assigned but is never used hence it is
and can be removed.

Cleans up clang warning:
warning: variable 'old_vlan' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqed: remove redundant pointer 'name'
Colin Ian King [Wed, 4 Jul 2018 12:06:26 +0000 (13:06 +0100)]
qed: remove redundant pointer 'name'

Pointer 'name' is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'name' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoethernet: micrel: remove redundant pointer 'info'
Colin Ian King [Wed, 4 Jul 2018 11:20:44 +0000 (12:20 +0100)]
ethernet: micrel: remove redundant pointer 'info'

Pointer 'info' is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'info' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hinic: remove redundant pointer pfhwdev
Colin Ian King [Wed, 4 Jul 2018 08:06:27 +0000 (09:06 +0100)]
net: hinic: remove redundant pointer pfhwdev

Pointer pfhwdev is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'pfhwdev' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: remove redundant variable 'protocol'
Colin Ian King [Wed, 4 Jul 2018 07:59:25 +0000 (08:59 +0100)]
net: hns3: remove redundant variable 'protocol'

Variable 'protocol' is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'protocol' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ethernet: gianfar_ethtool: remove redundant variable last_rule_idx
Colin Ian King [Wed, 4 Jul 2018 07:54:55 +0000 (08:54 +0100)]
net: ethernet: gianfar_ethtool: remove redundant variable last_rule_idx

Variable last_rule_idx is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'last_rule_idx' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: fec: remove redundant variable 'inc'
Colin Ian King [Wed, 4 Jul 2018 07:49:43 +0000 (08:49 +0100)]
net: fec: remove redundant variable 'inc'

Variable 'inc' is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'inc' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Fugang Duan <fugang.duan@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agocnic: remove redundant pointer req and variable func
Colin Ian King [Wed, 4 Jul 2018 07:39:12 +0000 (08:39 +0100)]
cnic: remove redundant pointer req and variable func

Pointer req and variable func are being assigned but are never used
hence they are redundant and can be removed.

Cleans up clang warnings:
warning: variable 'req' set but not used [-Wunused-but-set-variable]
warning: variable 'func' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: bgmac: remove redundant variable 'freed'
Colin Ian King [Wed, 4 Jul 2018 07:30:43 +0000 (08:30 +0100)]
net: bgmac: remove redundant variable 'freed'

Variable 'freed' is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'freed' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ethernet: nb8800: remove redundant pointer rxd
Colin Ian King [Wed, 4 Jul 2018 07:20:20 +0000 (08:20 +0100)]
net: ethernet: nb8800: remove redundant pointer rxd

Pointer rxd is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'rxb' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: alx: remove redundant variable old_duplex
Colin Ian King [Wed, 4 Jul 2018 07:15:30 +0000 (08:15 +0100)]
net: alx: remove redundant variable old_duplex

Variable old_duplex is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'old_duplex' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: alteon: acenic: remove redundant pointer rxdesc
Colin Ian King [Wed, 4 Jul 2018 07:01:35 +0000 (08:01 +0100)]
net: alteon: acenic: remove redundant pointer rxdesc

Pointer rxdesc is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'rxdesc' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: bcm_sf2: remove redundant variable off
Colin Ian King [Wed, 4 Jul 2018 06:54:36 +0000 (07:54 +0100)]
net: dsa: bcm_sf2: remove redundant variable off

Variable 'off' is being assigned but is never used hence it is
redundant and can be removed.

Cleans up clang warning:
warning: variable 'off' set but not used [-Wunused-but-set-variable]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'Scheduled-packet-Transmission-ETF'
David S. Miller [Wed, 4 Jul 2018 13:30:28 +0000 (22:30 +0900)]
Merge branch 'Scheduled-packet-Transmission-ETF'

Jesus Sanchez-Palencia says:

====================
Scheduled packet Transmission: ETF

Changes since v1:
  - moved struct sock_txtime from socket.h to uapi net_tstamp.h;
  - sk_clockid was changed from u16 to u8;
  - sk_txtime_flags was changed from u16 to a u8 bit field in struct sock;
  - the socket option flags are now validated in sock_setsockopt();
  - added SO_EE_ORIGIN_TXTIME;
  - sockc.transmit_time is now initialized from all IPv4 Tx paths;
  - added support for the IPv6 Tx path;

Overview
========

This work consists of a set of kernel interfaces that can be used by
applications that require (time-based) Scheduled Tx of packets.
It is comprised by 3 new components to the kernel:

  - SO_TXTIME: socket option + cmsg programming interfaces.

  - etf: the "earliest txtime first" qdisc, that provides per-queue
 TxTime-based scheduling. This has been renamed from 'tbs' to
 'etf' to better describe its functionality.

  - taprio: the "time-aware priority scheduler" qdisc, that provides
    per-port Time-Aware scheduling;

This patchset is providing the first 2 components, which have been
developed for longer. The taprio qdisc will be shared as an RFC separately
(shortly).

Note that this series is a follow up of the "Time based packet
transmission" RFCv3 [1].

etf (formerly known as 'tbs')
=============================

For applications/systems that the concept of time slices isn't precise
enough, the etf qdisc allows applications to control the instant when
a packet should leave the network controller. When used in conjunction
with taprio, it can also be used in case the application needs to
control with greater guarantee the offset into each time slice a packet
will be sent. Another use case of etf, is when only a small number of
applications on a system are time sensitive, so it can then be used
with a more traditional root qdisc (like mqprio).

The etf qdisc is designed so it buffers packets until a configurable
time before their deadline (Tx time). The qdisc uses a rbtree internally
so the buffered packets are always 'ordered' by their txtime (deadline)
and will be dequeued following the earliest txtime first.

It relies on the SO_TXTIME API set for receiving the per-packet timestamp
(txtime) as well as the config flags for each socket: the clockid to be
used as a reference, if the expected mode of txtime for that socket is
deadline or strict mode, and if packet drops should be reported on the
socket's error queue or not.

The qdisc will drop any packets with a Tx time in the past, or if a
packet expires while waiting for being dequeued. Drops can be reported
as errors back to userspace through the socket's error queue.

Example configuration:

$ tc qdisc add dev enp2s0 parent 100:1 etf offload delta 200000 \
            clockid CLOCK_TAI

Here, the Qdisc will use HW offload for the txtime control.
Packets will be dequeued by the qdisc "delta" (200000) nanoseconds before
their transmission time. Because this will be using HW offload and
since dynamic clocks are not supported by hrtimers, the system clock
and the PHC clock must be synchronized for this mode to behave as expected.

A more complete example can be found here, with instructions of how to
test it:

https://gist.github.com/jeez/bd3afeff081ba64a695008dd8215866f [2]

Note that we haven't modified the qdisc so it uses a timerqueue because
the modification needed was increasing the number of cachelines of a sk_buff.

This series is also hosted on github and can be found at [3].
The companion iproute2 patches can be found at [4].

[1] https://patchwork.ozlabs.org/cover/882342/

[2] github doesn't make it clear, but the gist can be cloned like this:
$ git clone https://gist.github.com/jeez/bd3afeff081ba64a695008dd8215866f scheduled-tx-tests

[3] https://github.com/jeez/linux/tree/etf-v2

[4] https://github.com/jeez/iproute2/tree/etf-v2
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/sched: Make etf report drops on error_queue
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:43:00 +0000 (15:43 -0700)]
net/sched: Make etf report drops on error_queue

Use the socket error queue for reporting dropped packets if the
socket has enabled that feature through the SO_TXTIME API.

Packets are dropped either on enqueue() if they aren't accepted by the
qdisc or on dequeue() if the system misses their deadline. Those are
reported as different errors so applications can react accordingly.

Userspace can retrieve the errors through the socket error queue and the
corresponding cmsg interfaces. A struct sock_extended_err* is used for
returning the error data, and the packet's timestamp can be retrieved by
adding both ee_data and ee_info fields as e.g.:

    ((__u64) serr->ee_data << 32) + serr->ee_info

This feature is disabled by default and must be explicitly enabled by
applications. Enabling it can bring some overhead for the Tx cycles
of the application.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigb: Add support for ETF offload
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:59 +0000 (15:42 -0700)]
igb: Add support for ETF offload

Implement HW offload support for SO_TXTIME through igb's Launchtime
feature. This is done by extending igb_setup_tc() so it supports
TC_SETUP_QDISC_ETF and configuring i210 so time based transmit
arbitration is enabled.

The FQTSS transmission mode added before is extended so strict
priority (SP) queues wait for stream reservation (SR) ones.
igb_config_tx_modes() is extended so it can support enabling/disabling
Launchtime following the previous approach used for the credit-based
shaper (CBS).

As the previous flow, FQTSS transmission mode is enabled automatically
by the driver once Launchtime (or CBS, as before) is enabled.
Similarly, it's automatically disabled when the feature is disabled
for the last queue that had it setup on.

The driver just consumes the transmit times from the skbuffs directly,
so no special handling is done in case an 'invalid' time is provided.
We assume this has been handled by the ETF qdisc already.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigb: Only call skb_tx_timestamp after descriptors are ready
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:58 +0000 (15:42 -0700)]
igb: Only call skb_tx_timestamp after descriptors are ready

Currently, skb_tx_timestamp() is being called before the Tx
descriptors are prepared in igb_xmit_frame_ring(), which happens
during either the igb_tso() or igb_tx_csum() calls.

Given that now the skb->tstamp might be used to carry the timestamp
for SO_TXTIME, we must only call skb_tx_timestamp() after the
information has been copied into the Tx descriptors.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigb: Refactor igb_offload_cbs()
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:57 +0000 (15:42 -0700)]
igb: Refactor igb_offload_cbs()

Split code into a separate function (igb_offload_apply()) that will be
used by ETF offload implementation.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigb: Only change Tx arbitration when CBS is on
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:56 +0000 (15:42 -0700)]
igb: Only change Tx arbitration when CBS is on

Currently the data transmission arbitration algorithm - DataTranARB
field on TQAVCTRL reg - is always set to CBS when the Tx mode is
changed from legacy to 'Qav' mode.

Make that configuration a bit more granular in preparation for the
upcoming Launchtime enabling patches, since CBS and Launchtime can be
enabled separately. That is achieved by moving the DataTranARB setup
to igb_config_tx_modes() instead.

Similarly, when disabling CBS we must check if it has been disabled
for all queues, and clear the DataTranARB accordingly.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigb: Refactor igb_configure_cbs()
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:55 +0000 (15:42 -0700)]
igb: Refactor igb_configure_cbs()

Make this function retrieve what it needs from the Tx ring being
addressed since it already relies on what had been saved on it before.
Also, since this function will be used by the upcoming Launchtime
patches rename it to better reflect its intention. Note that
Launchtime is not part of what 802.1Qav specifies, but the i210
datasheet refers to this set of functionality as "Qav Transmission
Mode".

Here we also perform a tiny refactor at is_any_cbs_enabled(), and add
further documentation to igb_setup_tx_mode().

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/sched: Add HW offloading capability to ETF
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:54 +0000 (15:42 -0700)]
net/sched: Add HW offloading capability to ETF

Add infra so etf qdisc supports HW offload of time-based transmission.

For hw offload, the time sorted list is still used, so packets are
dequeued always in order of txtime.

Example:

$ tc qdisc replace dev enp2s0 parent root handle 100 mqprio num_tc 3 \
           map 2 2 1 0 2 2 2 2 2 2 2 2 2 2 2 2 queues 1@0 1@1 2@2 hw 0

$ tc qdisc add dev enp2s0 parent 100:1 etf offload delta 100000 \
   clockid CLOCK_REALTIME

In this example, the Qdisc will use HW offload for the control of the
transmission time through the network adapter. The hrtimer used for
packets scheduling inside the qdisc will use the clockid CLOCK_REALTIME
as reference and packets leave the Qdisc "delta" (100000) nanoseconds
before their transmission time. Because this will be using HW offload and
since dynamic clocks are not supported by the hrtimer, the system clock
and the PHC clock must be synchronized for this mode to behave as
expected.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/sched: Introduce the ETF Qdisc
Vinicius Costa Gomes [Tue, 3 Jul 2018 22:42:53 +0000 (15:42 -0700)]
net/sched: Introduce the ETF Qdisc

The ETF (Earliest TxTime First) qdisc uses the information added
earlier in this series (the socket option SO_TXTIME and the new
role of sk_buff->tstamp) to schedule packets transmission based
on absolute time.

For some workloads, just bandwidth enforcement is not enough, and
precise control of the transmission of packets is necessary.

Example:

$ tc qdisc replace dev enp2s0 parent root handle 100 mqprio num_tc 3 \
           map 2 2 1 0 2 2 2 2 2 2 2 2 2 2 2 2 queues 1@0 1@1 2@2 hw 0

$ tc qdisc add dev enp2s0 parent 100:1 etf delta 100000 \
           clockid CLOCK_TAI

In this example, the Qdisc will provide SW best-effort for the control
of the transmission time to the network adapter, the time stamp in the
socket will be in reference to the clockid CLOCK_TAI and packets
will leave the qdisc "delta" (100000) nanoseconds before its transmission
time.

The ETF qdisc will buffer packets sorted by their txtime. It will drop
packets on enqueue() if their skbuff clockid does not match the clock
reference of the Qdisc. Moreover, on dequeue(), a packet will be dropped
if it expires while being enqueued.

The qdisc also supports the SO_TXTIME deadline mode. For this mode, it
will dequeue a packet as soon as possible and change the skb timestamp
to 'now' during etf_dequeue().

Note that both the qdisc's and the SO_TXTIME ABIs allow for a clockid
to be configured, but it's been decided that usage of CLOCK_TAI should
be enforced until we decide to allow for other clockids to be used.
The rationale here is that PTP times are usually in the TAI scale, thus
no other clocks should be necessary. For now, the qdisc will return
EINVAL if any clocks other than CLOCK_TAI are used.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/sched: Allow creating a Qdisc watchdog with other clocks
Vinicius Costa Gomes [Tue, 3 Jul 2018 22:42:52 +0000 (15:42 -0700)]
net/sched: Allow creating a Qdisc watchdog with other clocks

This adds 'qdisc_watchdog_init_clockid()' that allows a clockid to be
passed, this allows other time references to be used when scheduling
the Qdisc to run.

Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: packet: Hook into time based transmission.
Richard Cochran [Tue, 3 Jul 2018 22:42:51 +0000 (15:42 -0700)]
net: packet: Hook into time based transmission.

For raw layer-2 packets, copy the desired future transmit time from
the CMSG cookie into the skb.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ipv6: Hook into time based transmission
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:50 +0000 (15:42 -0700)]
net: ipv6: Hook into time based transmission

Add a struct sockcm_cookie parameter to ip6_setup_cork() so
we can easily re-use the transmit_time field from struct inet_cork
for most paths, by copying the timestamp from the CMSG cookie.
This is later copied into the skb during __ip6_make_skb().

For the raw fast path, also pass the sockcm_cookie as a parameter
so we can just perform the copy at rawv6_send_hdrinc() directly.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ipv4: Hook into time based transmission
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:49 +0000 (15:42 -0700)]
net: ipv4: Hook into time based transmission

Add a transmit_time field to struct inet_cork, then copy the
timestamp from the CMSG cookie at ip_setup_cork() so we can
safely copy it into the skb later during __ip_make_skb().

For the raw fast path, just perform the copy at raw_send_hdrinc().

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: Add a new socket option for a future transmit time.
Richard Cochran [Tue, 3 Jul 2018 22:42:48 +0000 (15:42 -0700)]
net: Add a new socket option for a future transmit time.

This patch introduces SO_TXTIME. User space enables this option in
order to pass a desired future transmit time in a CMSG when calling
sendmsg(2). The argument to this socket option is a 8-bytes long struct
provided by the uapi header net_tstamp.h defined as:

struct sock_txtime {
clockid_t  clockid;
u32 flags;
};

Note that new fields were added to struct sock by filling a 2-bytes
hole found in the struct. For that reason, neither the struct size or
number of cachelines were altered.

Signed-off-by: Richard Cochran <rcochran@linutronix.de>
Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: Clear skb->tstamp only on the forwarding path
Jesus Sanchez-Palencia [Tue, 3 Jul 2018 22:42:47 +0000 (15:42 -0700)]
net: Clear skb->tstamp only on the forwarding path

This is done in preparation for the upcoming time based transmission
patchset. Now that skb->tstamp will be used to hold packet's txtime,
we must ensure that it is being cleared when traversing namespaces.
Also, doing that from skb_scrub_packet() before the early return would
break our feature when tunnels are used.

Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoisdn: mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 3 Jul 2018 21:17:31 +0000 (16:17 -0500)]
isdn: mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Warning level 2 was used: -Wimplicit-fallthrough=2

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: usb: asix: allow optionally getting mac address from device tree
Marcel Ziswiler [Tue, 3 Jul 2018 15:06:49 +0000 (17:06 +0200)]
net: usb: asix: allow optionally getting mac address from device tree

For Embedded use where e.g. AX88772B chips may be used without external
EEPROMs the boot loader may choose to pass the MAC address to be used
via device tree. Therefore, allow for optionally getting the MAC
address from device tree data e.g. as follows (excerpt from a T30 based
board, local-mac-address to be filled in by boot loader):

/* EHCI instance 1: USB2_DP/N -> AX88772B */
usb@7d004000 {
status = "okay";
#address-cells = <1>;
#size-cells = <0>;
asix@1 {
reg = <1>;
local-mac-address = [00 00 00 00 00 00];
};
};

Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_pedit: fix possible memory leak in tcf_pedit_init()
Wei Yongjun [Tue, 3 Jul 2018 13:45:12 +0000 (13:45 +0000)]
net: sched: act_pedit: fix possible memory leak in tcf_pedit_init()

'keys_ex' is malloced by tcf_pedit_keys_ex_parse() in tcf_pedit_init()
but not all of the error handle path free it, this may cause memory
leak. This patch fix it.

Fixes: 71d0ed7079df ("net/act_pedit: Support using offset relative to the conventional network headers")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'bridge-iproute2-isolated-port-and-selftests'
David S. Miller [Wed, 4 Jul 2018 12:40:02 +0000 (21:40 +0900)]
Merge branch 'bridge-iproute2-isolated-port-and-selftests'

Nikolay Aleksandrov says:

====================
bridge: iproute2 isolated port and selftests

Add support to iproute2 for port isolation config and selftests for it.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoselftests: forwarding: test for bridge port isolation
Nikolay Aleksandrov [Tue, 3 Jul 2018 12:42:44 +0000 (15:42 +0300)]
selftests: forwarding: test for bridge port isolation

This test checks if the bridge port isolation feature works as expected
by performing ping/ping6 tests between hosts that are isolated (should
not work) and between an isolated and non-isolated hosts (should work).
Same test is performed for flooding from and to isolated and
non-isolated ports.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoselftests: forwarding: lib: extract ping and ping6 so they can be reused
Nikolay Aleksandrov [Tue, 3 Jul 2018 12:42:43 +0000 (15:42 +0300)]
selftests: forwarding: lib: extract ping and ping6 so they can be reused

Extract ping and ping6 command execution so the return value can be
checked by the caller, this is needed for port isolation tests that are
intended to fail.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'vhost_net-Avoid-vq-kicks-during-busyloop'
David S. Miller [Wed, 4 Jul 2018 12:30:47 +0000 (21:30 +0900)]
Merge branch 'vhost_net-Avoid-vq-kicks-during-busyloop'

Toshiaki Makita says:

====================
vhost_net: Avoid vq kicks during busyloop

Under heavy load vhost tx busypoll tend not to suppress vq kicks, which
causes poor guest tx performance. The detailed scenario is described in
commitlog of patch 2.
Rx seems not to have that serious problem, but for consistency I made a
similar change on rx to avoid rx wakeups (patch 3).
Additionary patch 4 is to avoid rx kicks under heavy load during
busypoll.

Tx performance is greatly improved by this change. I don't see notable
performance change on rx with this series though.

Performance numbers (tx):

- Bulk transfer from guest to external physical server.
    [Guest]->vhost_net->tap--(XDP_REDIRECT)-->i40e --(wire)--> [Server]
- Set 10us busypoll.
- Guest disables checksum and TSO because of host XDP.
- Measured single flow Mbps by netperf, and kicks by perf kvm stat
  (EPT_MISCONFIG event).

                            Before              After
                          Mbps  kicks/s      Mbps  kicks/s
UDP_STREAM 1472byte              247758                 27
                Send   3645.37            6958.10
                Recv   3588.56            6958.10
              1byte                9865                 37
                Send      4.34               5.43
                Recv      4.17               5.26
TCP_STREAM             8801.03    45794   9592.77     2884

v2:
- Split patches into 3 parts (renaming variables, tx-kick fix, rx-wakeup
  fix).
- Avoid rx-kicks too (patch 4).
- Don't memorize endtime as it is not needed for now.
====================

Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agovhost_net: Avoid rx vring kicks during busyloop
Toshiaki Makita [Tue, 3 Jul 2018 07:31:34 +0000 (16:31 +0900)]
vhost_net: Avoid rx vring kicks during busyloop

We may run out of avail rx ring descriptor under heavy load but busypoll
did not detect it so busypoll may have exited prematurely. Avoid this by
checking rx ring full during busypoll.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agovhost_net: Avoid rx queue wake-ups during busypoll
Toshiaki Makita [Tue, 3 Jul 2018 07:31:33 +0000 (16:31 +0900)]
vhost_net: Avoid rx queue wake-ups during busypoll

We may run handle_rx() while rx work is queued. For example a packet can
push the rx work during the window before handle_rx calls
vhost_net_disable_vq().
In that case busypoll immediately exits due to vhost_has_work()
condition and enables vq again. This can lead to another unnecessary rx
wake-ups, so poll rx work instead of enabling the vq.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agovhost_net: Avoid tx vring kicks during busyloop
Toshiaki Makita [Tue, 3 Jul 2018 07:31:32 +0000 (16:31 +0900)]
vhost_net: Avoid tx vring kicks during busyloop

Under heavy load vhost busypoll may run without suppressing
notification. For example tx zerocopy callback can push tx work while
handle_tx() is running, then busyloop exits due to vhost_has_work()
condition and enables notification but immediately reenters handle_tx()
because the pushed work was tx. In this case handle_tx() tries to
disable notification again, but when using event_idx it by design
cannot. Then busyloop will run without suppressing notification.
Another example is the case where handle_tx() tries to enable
notification but avail idx is advanced so disables it again. This case
also leads to the same situation with event_idx.

The problem is that once we enter this situation busyloop does not work
under heavy load for considerable amount of time, because notification
is likely to happen during busyloop and handle_tx() immediately enables
notification after notification happens. Specifically busyloop detects
notification by vhost_has_work() and then handle_tx() calls
vhost_enable_notify(). Because the detected work was the tx work, it
enters handle_tx(), and enters busyloop without suppression again.
This is likely to be repeated, so with event_idx we are almost not able
to suppress notification in this case.

To fix this, poll the work instead of enabling notification when
busypoll is interrupted by something. IMHO vhost_has_work() is kind of
interruption rather than a signal to completely cancel the busypoll, so
let's run busypoll after the necessary work is done.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agovhost_net: Rename local variables in vhost_net_rx_peek_head_len
Toshiaki Makita [Tue, 3 Jul 2018 07:31:31 +0000 (16:31 +0900)]
vhost_net: Rename local variables in vhost_net_rx_peek_head_len

So we can easily see which variable is for which, tx or rx.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>